Automatic segmentation method for full-mouth crown based on gum line identification and region growing

Abstract

The application discloses a full-mouth tooth crown automatic segmentation method based on gum line identification and region growth, relates to the field of dental digital restoration, and comprises the following steps: pre-processing a tooth model obtained by mouth scanning to obtain an initial mesh model; identifying a gum line based on the initial mesh model to obtain gum line region mesh nodes; segmenting a gum and a tooth crown region based on the gum line region mesh nodes to obtain a tooth crown region mesh model; and segmenting the tooth crown region mesh model to obtain mesh models of individual tooth crowns. The application provides the full-mouth tooth crown automatic segmentation method based on the gum line identification and the region growth, so as to solve the problems of inaccurate gum line identification, poor region growth segmentation effect and poor full-mouth adaptability in the prior art, realize accurate extraction and feature reservation of the full-mouth tooth crown, and provide reliable support for automation and standardization of digital oral restoration.

Description

Technical Field

[0001] This invention relates to the field of digital dental restoration, specifically to a method for automatic segmentation of full-mouth crowns based on gingival line recognition and regional growth. Background Technology

[0002] In the field of prosthodontics, precise segmentation of full-mouth crowns is the foundation for subsequent prosthodontic design (such as crown fit, orthodontic treatment plan formulation, and implant guide fabrication). Its precision directly determines the clinical outcome of prosthodontics (such as marginal fit of the prosthodontic restoration and recovery of the patient's chewing function).

[0003] Traditional full-mouth crown segmentation is mostly completed by manually marking the gingival line and separating the crowns one by one by the dentist. It relies entirely on the operator's clinical experience and has the following drawbacks: (1) High subjectivity: Different dentists mark different gingival lines, resulting in inconsistent segmentation results; (2) Low efficiency: Full-mouth processing takes a long time and cannot meet the needs of clinical batch processing; (3) Incomplete feature preservation: Key structures such as cusps and marginal ridges are easily deleted due to manual operation; (4) Insufficient standardization: It is impossible to guarantee the consistency of segmentation results in different cases. These problems seriously restrict the automated and large-scale development of digital dental restoration and have become a bottleneck for the clinical popularization of digital technology.

[0004] With the advancement of digital dental technology, automatic crown segmentation algorithms (such as threshold segmentation, edge detection, and region growth) have been proposed to replace manual operations. However, existing automatic crown segmentation technologies still have significant drawbacks: (1) Inaccurate gingival line recognition: Affected by factors such as gingival swelling, scanning noise, and crowded dentition, it is impossible to accurately locate the boundary between the gingiva and the crown, which easily leads to large errors; (2) Poor region growth segmentation effect: Seed point selection depends on human experience, often resulting in oversegmentation (misjudging adjacent crowns as a region) or undersegmentation (missing some crown structures); (3) Poor adaptability to the whole mouth: Most are designed for single teeth and cannot efficiently handle the complex morphology of the entire dentition, making it difficult to meet the needs of full-mouth restoration. These shortcomings make it difficult for existing automatic segmentation technologies to meet the clinical requirements of "precision, efficiency, and consistency," limiting their widespread application in digital dental restoration. Summary of the Invention

[0005] This invention provides an automatic segmentation method for full-mouth crowns based on gingival line recognition and regional growth, in order to solve the problems of inaccurate gingival line recognition, poor regional growth segmentation effect, and poor adaptability to full mouth in the prior art. It achieves accurate extraction and feature preservation of full-mouth crowns, and provides reliable support for the automation and standardization of digital oral restoration.

[0006] This invention is achieved through the following technical solution:

[0007] An automated segmentation method for full-mouth crowns based on gingival line recognition and regional growth includes the following steps:

[0008] S1. Preprocess the dental arch model obtained from the oral cavity scan to obtain the initial mesh model;

[0009] S2. Identify the gingival line based on the initial mesh model to obtain the mesh nodes of the gingival line region;

[0010] S3. Based on the gingival line region mesh nodes, the gingival and crown regions are segmented to obtain the crown region mesh model;

[0011] S4. Divide the tooth crown region mesh model to obtain the mesh model of each individual tooth crown.

[0012] To address the problems of inaccurate gingival line recognition, poor regional growth segmentation, and poor adaptability to the entire mouth in existing technologies, this invention proposes an automatic full-mouth crown segmentation method based on gingival line recognition and regional growth. This method first obtains a dental arch model using existing intraoral scanning technology, and after preprocessing, obtains an initial mesh model. Then, based on the preprocessed initial mesh model, gingival line region mesh nodes are obtained using gingival line recognition technology. Next, based on the gingival line region mesh nodes, the gingiva and crown are segmented to obtain a crown region mesh model. The crown region mesh model obtained in this application is still a single unit; therefore, it is necessary to further segment the crown region mesh model to finally obtain the mesh models of each individual crown.

[0013] Compared with existing automatic crown segmentation technologies, this application can overcome the technical defects of existing technologies, such as large gingival line recognition errors, poor regional growth segmentation effects, and low full-mouth processing efficiency. It truly achieves accurate extraction and feature preservation of full-mouth crowns, providing reliable basic data support for the automation and standardization of digital oral restoration, and is conducive to the promotion and application of automatic segmentation technology in digital oral restoration.

[0014] Furthermore, step S1 specifically includes:

[0015] S101. Retain all the crowns and the gingiva connected to the crowns in the dental arch model obtained by oral scanning to obtain the first mesh model;

[0016] S102. Perform coordinate registration on the first mesh model so that the gingival bottom surface of the first mesh model is coplanar with the XY plane, and the Z-axis extends upward and toward the direction of the crown, thus obtaining the second mesh model.

[0017] S103. Perform adaptive smoothing on the second mesh model to obtain the initial mesh model.

[0018] Those skilled in the art should understand that the dental arch model obtained from intraoral scanning includes data from the crown region, gingival region, and other invalid regions. Before formal processing, this method performs a coarse removal of the gingival region and other regions from the dental arch model to reduce subsequent computational load and difficulty, ultimately retaining only the entire crown and the portion of the gingival region adjacent to the crown. Then, the retained model is coordinate-registered for subsequent processing: the bottom of the gingival region of the model is made coplanar with the XY plane, and the Z-axis points upward and extends towards the direction of the crown, thus obtaining the second mesh model. The surface of the second mesh model may have irregular protrusions or burrs due to scanning noise. To obtain a smoother crown model that retains important features (such as cusps and marginal ridges), this method performs adaptive smoothing on the second mesh model to smooth noise and preserve important geometric features of the crown.

[0019] In addition, this scheme makes the gingival bottom surface of the first mesh model coplanar with the XY plane, and sets the Z-axis to extend upwards and toward the direction of the crown, so it can be used in the maxillary or mandibular dentition and has strong versatility.

[0020] Furthermore, step S2 specifically includes:

[0021] S201. Calculate the average curvature of all grid nodes in the initial grid model;

[0022] S202. Filter the grid nodes with the minimum average curvature within their neighborhood to obtain the first point set. For example, if grid node A has an average curvature of X, determine if there are other grid nodes with an average curvature less than X within its neighborhood radius. If not, place grid node A in the first point set; otherwise, grid node A does not belong to the first point set. The purpose of this step is to obtain a set of points containing grid nodes in the gingival line region, as a preliminary preparation for finally locking the grid nodes in the gingival line region.

[0023] S203. Select grid nodes with an average curvature less than the first threshold from the initial grid model to obtain the second point set;

[0024] S204. Calculate the average angle of the normal vectors of all grid nodes in the neighborhood in the initial grid model; filter the grid nodes whose average angle of the normal vectors is greater than the second threshold to obtain the third point set;

[0025] S205. Set the coordinate range of the gingival line in the Z-axis direction, filter the grid nodes located in the coordinate range, and obtain the fourth point set;

[0026] S206. Take the intersection of the first point set, the second point set, the third point set, and the fourth point set to obtain the fifth point set;

[0027] S207. Perform clustering processing on the fifth point set to remove outliers and noise points, and obtain the gingival line region mesh nodes.

[0028] This method for obtaining the gingival line region mesh nodes employs multi-scale feature fusion gingival line recognition technology. Through multiple techniques, including average curvature, average angle of normal vectors, comprehensive screening of Z-axis coordinate intervals, and clustering, it solves the problems of large gingival line positioning errors and inaccurate boundary recognition caused by gingival swelling, crowded teeth, and scanning noise in existing technologies. This significantly improves the accuracy of locating the boundary between the gingiva and the crown, and enhances the accuracy of gingival line recognition.

[0029] Furthermore, in step S201, the average curvature of any grid node is calculated using the following formula:

[0030] ;

[0031] In the formula: H is the average curvature; E, F, and G are the first basic form coefficients corresponding to the grid node; L, M, and N are the second basic form coefficients corresponding to the grid node;

[0032] In step S204, the average angle between the normal vectors of any grid node in its neighborhood is calculated using the following formula:

[0033] ;

[0034] In the formula: The average angle between the normal vectors of node i in its neighborhood; N(i) represents the set of neighboring nodes of node i; |N(i)| is the number of neighboring nodes; j represents the j-th neighboring node; The normal vector representing node i; This represents the normal vector of the j-th neighboring node.

[0035] Those skilled in the art should understand that the first and second basic forms are existing techniques for studying surface properties, and the corresponding first and second basic form coefficients are also existing technical terms. This solution uses the first and second basic form coefficients to calculate the average curvature of the mesh nodes.

[0036] Furthermore, step S3 specifically includes:

[0037] S301. Using the region growing algorithm, the grid node with the largest Z-axis coordinate is selected as the first seed point on the initial grid model;

[0038] S302. Using the first seed point as a reference, determine the boundary zone between the gingiva and the crown based on the gingival line region mesh nodes;

[0039] S303. Based on the grid nodes within the boundary band, the gingival and crown regions are segmented to obtain a grid model of the crown region.

[0040] In step S2, the gingival line region mesh nodes have been obtained, but these mesh nodes are essentially discrete points and cannot generate a good and continuous gingival line for accurate segmentation of the gingiva and crown. To overcome this problem, this scheme uses a region growing algorithm to accurately extract the boundary zone between the gingiva and crown, thereby accurately extracting the crown region.

[0041] Furthermore, step S302 specifically includes:

[0042] S3021. Starting from the first seed point, the grid grows outwards to the surrounding grid nodes, and all boundary nodes are determined using the following method:

[0043] If the distance from a certain grid node to all grid nodes in the gingival region is greater than or equal to the third threshold, then the grid node is marked as a crown and continues to grow towards the surrounding grid nodes;

[0044] If the distance from a certain grid node to any grid node in the gingival region is less than the third threshold, then the grid node is marked as a boundary node and growth stops;

[0045] S3022. Starting from each boundary node, extend the grid to the unvisited grid nodes by a specified number of layers, and mark all grid nodes within the extension area as boundary nodes;

[0046] S3023. The set of all grid nodes marked as boundary nodes is used as the boundary zone between the gingiva and the crown.

[0047] The purpose of introducing the "boundary band" in this application is to improve the segmentation accuracy of the crown region as much as possible. Therefore, this scheme uses the idea of ​​distance constraint, combined with the previously determined gingival line region mesh nodes, to obtain all the boundary nodes. Then, based on each boundary node, it extends to the mesh nodes that the region growth algorithm has not yet visited, thereby obtaining complete boundary nodes, and collecting all the boundary nodes as the boundary band.

[0048] In this scheme, the specified number of layers can be adaptively set according to the actual situation, and no specific limitation is made here.

[0049] Furthermore, step S303 specifically includes:

[0050] S3031. Within the boundary zone between the gingiva and the crown, select the grid node with the largest Z-axis coordinate as the second seed point;

[0051] S3032. Within the boundary zone between the gingiva and the crown, starting from the second seed point, the mesh nodes grow outwards to the surrounding mesh nodes, and the mesh nodes within the boundary zone are divided using the following method:

[0052] If the average curvature of a certain grid node is greater than or equal to the fourth threshold, then the grid node is marked as a crown and continues to grow towards the surrounding grid nodes;

[0053] If the average curvature of a grid node is less than the fourth threshold, the grid node is marked as non-crown and growth is stopped.

[0054] S3033. Combine all the mesh nodes marked as crowns to obtain the crown region mesh model.

[0055] This scheme employs a region growing algorithm and introduces curvature constraints. These constraints are used to perform a secondary screening of all boundary nodes within the boundary band, aiming to retain as many mesh nodes as possible from the crown region. Mesh nodes within the boundary band not marked as crown nodes are assumed to belong to the gingival region mesh model.

[0056] As can be seen, the process of obtaining the crown region mesh model in this application adopts a two-stage region growth method of first distance constraint and then curvature constraint, combined with the boundary band propagation mechanism, which solves the problems of seed point selection relying on manual experience and easy over-segmentation / under-segmentation in traditional region growth technology, realizes the complete automated extraction of the crown region, and significantly improves the extraction accuracy of the entire crown.

[0057] Furthermore, in step S4, the method for segmenting the tooth crown region mesh model includes:

[0058] S401. Calculate the average angle between the normal vectors of all grid nodes in the neighborhood within the mesh model of the crown region;

[0059] S402. Based on the average angle between the normal vectors of all grid nodes in their neighborhood, filter the grid nodes in the gap region to obtain the sixth point set.

[0060] S403. Perform clustering on the sixth point set and retain the top N clusters with the most nodes; where N = number of teeth - 1;

[0061] S404. Perform plane fitting on the mesh nodes in each of the N clusters to obtain N cutting planes;

[0062] S405. The tooth crown region mesh model is divided by N cutting planes to obtain (N+1) tooth crown mesh models, and the boundary of the mesh model of each tooth crown is smoothed.

[0063] In this application, the crown region mesh model extracted in step S3 is still a whole. To obtain a single crown model, this scheme uses the average angle of the normal vectors of the mesh nodes to identify the mesh nodes at the interdental spaces. After clustering and plane fitting, N cutting planes are obtained, thereby segmenting the crown region mesh model to obtain a single crown model. The edges of the single crown mesh model have jagged irregular patches, so the boundaries are smoothed.

[0064] This solution achieves one-time batch separation of full-mouth crowns by using threshold recognition of the angle between the normal vectors of interdental nodes and plane fitting segmentation technology, solving the problem that existing technologies only support single-tooth processing and have poor adaptability to full-mouth conditions. Furthermore, by employing an edge optimization approach of smoothing after cutting, it outputs single-crown models that meet clinical restoration requirements, addressing the issues of rough segmentation boundaries and the need for secondary manual refining in existing methods, thus promoting the standardized development of digital dental restoration.

[0065] Furthermore, in step S402, the method for filtering grid nodes in the interdental region is as follows: if the average angle of the normal vectors of a certain grid node in its neighborhood is greater than the fifth threshold, then it is marked as a grid node in the interdental region.

[0066] Furthermore, the clustering process in step S403 uses the DBSCAN clustering algorithm; the plane fitting in step S404 uses a plane fitting algorithm based on principal component analysis.

[0067] It should be noted that, since each person's dental arch appearance is different, the first threshold, second threshold, third threshold, fourth threshold and fifth threshold, as well as the Z-axis coordinate range in this application are not specifically limited here. When using this method, they should be adaptively set according to the specific dental arch appearance.

[0068] Furthermore, if the method of this application is required to segment the crowns of the maxillary and mandibular dentitions, it is best to perform the method of this application separately for the maxillary and mandibular dentitions to ensure the accuracy of segmentation.

[0069] Compared with the prior art, the present invention has at least the following advantages and beneficial effects:

[0070] 1. The present invention is an automatic segmentation method for full-mouth crowns based on gingival line recognition and region growth. It can overcome the technical defects of existing technologies, such as large gingival line recognition error, poor region growth segmentation effect, and low full-mouth processing efficiency. It truly realizes the accurate extraction and feature preservation of full-mouth crowns, providing reliable basic data support for the automation and standardization of digital oral restoration, and is conducive to the promotion and application of automatic segmentation technology in digital oral restoration.

[0071] 2. This invention provides an automatic full-mouth crown segmentation method based on gingival line recognition and region growth. It employs multi-scale feature fusion gingival line recognition technology to obtain gingival line region mesh nodes. Through multiple technical means such as average curvature, average angle of normal vector, Z-axis coordinate interval comprehensive screening, and clustering processing, it solves the problems of large gingival line positioning errors and inaccurate boundary recognition caused by gingival swelling, crowded teeth, and scanning noise in existing technologies. This significantly improves the accuracy of locating the boundary between the gingiva and the crown, and enhances the accuracy of gingival line recognition.

[0072] 3. The present invention is an automatic segmentation method for full-mouth crowns based on gingival line recognition and region growth. It adopts a two-stage region growth method of first distance constraint and then curvature constraint to obtain the crown region mesh model. Combined with the boundary band propagation mechanism, it solves the problems of seed point selection relying on human experience and easy over-segmentation / under-segmentation in traditional region growth technology. It realizes complete and automated extraction of the crown region and significantly improves the extraction accuracy of full-mouth crowns.

[0073] 4. This invention provides an automatic full-mouth crown segmentation method based on gingival line recognition and region growth. Through threshold recognition of the angle between the normal vectors of interdental nodes and planar fitting segmentation technology, it achieves one-time batch separation of full-mouth crowns, solving the problems of existing technologies that only support single-tooth processing and have poor adaptability to full-mouth applications. Furthermore, by employing an edge optimization approach of smoothing after cutting, it outputs single-crown models that meet clinical restoration requirements, addressing the issues of rough segmentation boundaries and the need for secondary manual refining in existing methods. This is conducive to the standardized development of digital dental restoration. Attached Figure Description

[0074] The accompanying drawings, which are included to provide a further understanding of embodiments of the invention and form part of this application, do not constitute a limitation thereof. In the drawings:

[0075] Figure 1 This is a flowchart illustrating a specific embodiment of the present invention;

[0076] Figure 2 This is a schematic diagram of the first mesh model in a specific embodiment of the present invention;

[0077] Figure 3 This is a schematic diagram of the second mesh model in a specific embodiment of the present invention;

[0078] Figure 4 This is a schematic diagram of the initial mesh model in a specific embodiment of the present invention;

[0079] Figure 5 This is a schematic diagram of the first point set in a specific embodiment of the present invention;

[0080] Figure 6This is a schematic diagram of the second point set in a specific embodiment of the present invention;

[0081] Figure 7 This is a schematic diagram of the third point set in a specific embodiment of the present invention;

[0082] Figure 8 This is a schematic diagram of the fifth point set in a specific embodiment of the present invention;

[0083] Figure 9 This is a schematic diagram of the grid nodes in the gingival line region in a specific embodiment of the present invention;

[0084] Figure 10 This is a schematic diagram of the boundary band in a specific embodiment of the present invention;

[0085] Figure 11 This is a schematic diagram of the full crown area in a specific embodiment of the present invention;

[0086] Figure 12 This is a schematic diagram of the mesh nodes in the interdental region in a specific embodiment of the present invention;

[0087] Figure 13 This is a schematic diagram of the cutting plane in a specific embodiment of the present invention;

[0088] Figure 14 This is a segmented tooth crown mesh model in a specific embodiment of the present invention;

[0089] Figure 15 This is the final tooth crown mesh model obtained in a specific embodiment of the present invention. Detailed Implementation

[0090] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the embodiments and accompanying drawings. The illustrative embodiments and descriptions of this invention are only for explaining this invention and are not intended to limit this invention.

[0091] Example 1:

[0092] like Figure 1 The automatic segmentation method for full-mouth crowns based on gingival line recognition and region growth is illustrated in this embodiment using the maxillary dentition of a patient as an example. The method includes the following steps:

[0093] Step S1: Preprocess the dental arch model obtained from the oral scan to obtain the initial mesh model.

[0094] The specific process is as follows:

[0095] S101. Retain all crowns and the portion of gingiva connecting to the crowns in the dental arch model obtained by oral scanning to obtain the first mesh model, such as... Figure 2 As shown.

[0096] S102. Perform coordinate registration on the first mesh model, making the gingival base of the first mesh model coplanar with the XY plane, and extending the Z-axis upwards and towards the direction of the crown, to obtain the second mesh model, as shown below. Figure 3 As shown.

[0097] S103. Adaptive smoothing is applied to the second mesh model to obtain the initial mesh model. In this embodiment, an anisotropic mesh smoothing algorithm is used for adaptive smoothing. This algorithm automatically adjusts the smoothing intensity based on local curvature, and in this method, it smooths noise while preserving the geometric features of the teeth to the maximum extent. The initial mesh model obtained in this embodiment is as follows: Figure 4 As shown.

[0098] Step S2: Identify the gingival line based on the initial mesh model to obtain the mesh nodes of the gingival line region.

[0099] The specific process is as follows:

[0100] S201. Calculate the average curvature of all grid nodes in the initial grid model;

[0101] For any grid node, its average curvature is calculated using the following formula:

[0102] ;

[0103] In the formula: H is the average curvature; E, F, and G are the first basic form coefficients corresponding to the grid node; L, M, and N are the second basic form coefficients corresponding to the grid node.

[0104] S202. Filter the grid nodes with the minimum average curvature in the neighborhood to obtain the first point set.

[0105] For example, if the average curvature of grid node A is X, determine whether there are other grid nodes with an average curvature less than X within its neighborhood radius; if not, place grid node A in the first point set; if so, grid node A does not belong to the first point set.

[0106] In this embodiment, the neighborhood radius is set to 0.5 mm; the resulting first point set is as follows: Figure 5 As shown.

[0107] S203. In the initial mesh model, mesh nodes with an average curvature less than a first threshold are selected to obtain a second point set; in this embodiment, the first threshold is -0.3mm. -1 ~-0.2mm -1 Preferably -0.25mm -1 The second set of points obtained is as follows Figure 6 As shown.

[0108] S204. Calculate the average angle between the normal vectors of all grid nodes in the neighborhood of the initial grid model.

[0109] For any grid node, the average angle between the normal vectors in its neighborhood is calculated using the following formula:

[0110] ;

[0111] In the formula: The average angle between the normal vectors of node i in its neighborhood; N(i) represents the set of neighboring nodes of node i; |N(i)| is the number of neighboring nodes; j represents the j-th neighboring node; The normal vector representing node i; This represents the normal vector of the j-th neighboring node.

[0112] Grid nodes with an average angle between their normal vectors greater than a second threshold are selected to obtain a third set of points. In this embodiment, the second threshold is 0.15~0.25 rad; preferably 0.2 rad. The resulting third set of points is as follows: Figure 7 As shown.

[0113] S205. Set the coordinate range of the gingival line in the Z-axis direction, filter the grid nodes located in the coordinate range, and obtain the fourth point set; in this embodiment, the coordinate range in the Z-axis direction is [0.5, 6].

[0114] S206. Take the intersection of the first, second, third, and fourth point sets to obtain the fifth point set, as follows: Figure 8 As shown.

[0115] S207. Perform clustering on the fifth point set to remove outliers and noise points, obtaining the gingival line region mesh nodes. Please refer to... Figure 9 : Figure 9 The red grid nodes are noise points, the gray grid nodes are outliers, and the green grid nodes are the final grid nodes for the gingival line region.

[0116] Step S3: Based on the gingival line region mesh nodes, segment the gingival and crown regions to obtain the crown region mesh model.

[0117] The specific process is as follows:

[0118] S301. Select the grid node with the largest Z-axis coordinate on the initial grid model as the first seed point.

[0119] S302. Using a region growing algorithm, with the first seed point as a reference, and based on the gingival line region mesh nodes, determine the boundary zone between the gingiva and the crown. Specific methods include:

[0120] S3021. Using a region growing algorithm, the region grows from the first seed point outwards to the surrounding grid nodes, and all boundary nodes are determined using the following method:

[0121] If the distance from a certain grid node to all grid nodes in the gingival region is greater than or equal to the third threshold, then the grid node is marked as a crown and continues to grow towards the surrounding grid nodes;

[0122] If the distance from a certain grid node to any grid node in the gingival region is less than the third threshold, then the grid node is marked as a boundary node and growth stops.

[0123] In this embodiment, the third threshold is set to 2.5~3mm, preferably 2.8mm.

[0124] S3022. Starting from each boundary node, extend the grid to the unvisited grid nodes by a specified number of layers, and mark all grid nodes within the extension area as boundary nodes; in this embodiment, the specified number of layers is 5 layers.

[0125] S3023. The set of all mesh nodes marked as boundary nodes is taken as the boundary band between the gingiva and the crown. The boundary band extracted in this embodiment is as follows: Figure 10 As shown, Figure 10 The yellow parts represent the boundary zone, and the red parts represent the grid nodes of the gingival line area.

[0126] S303. Based on the mesh nodes within the boundary band, segment the gingival and crown regions to obtain a mesh model of the crown region. Specifically, this includes:

[0127] S3031. Within the boundary zone between the gingiva and the crown, select the grid node with the largest Z-axis coordinate as the second seed point;

[0128] S3032. Within the boundary zone between the gingiva and the crown, starting from the second seed point, the mesh nodes grow outwards to the surrounding mesh nodes, and the mesh nodes within the boundary zone are divided using the following method:

[0129] If the average curvature of a certain grid node is greater than or equal to the fourth threshold, then the grid node is marked as a crown and continues to grow towards the surrounding grid nodes;

[0130] If the average curvature of a grid node is less than the fourth threshold, the grid node is marked as non-crown and growth is stopped.

[0131] In this embodiment, the fourth threshold is -0.3mm. -1 ~-0.2mm -1 Preferably -0.25mm -1 .

[0132] S3033. Combine all mesh nodes marked as crowns to obtain the crown region mesh model as follows: Figure 11 As shown in the green section.

[0133] Step S4: Divide the tooth crown region mesh model to obtain the mesh model of each individual tooth crown.

[0134] Example 2:

[0135] An automatic segmentation method for full-mouth crowns based on gingival line recognition and region growth, building upon Example 1, includes the following steps for the crown region mesh model method:

[0136] S401. Calculate the average angle between the normal vectors of all grid nodes in the neighborhood within the grid model of the crown region; for specific calculation methods, please refer to step S204.

[0137] S402. Based on the average angle of the normal vectors of all grid nodes in their neighborhood, filter the grid nodes in the interdental region to obtain the sixth point set. The filtering method is as follows: if the average angle of the normal vectors of a certain grid node in its neighborhood is greater than the fifth threshold, then mark it as a grid node in the interdental region.

[0138] In this embodiment, the fifth threshold is 0.5~0.7 rad; preferably 0.6 rad, and the resulting interdental region mesh nodes are as follows: Figure 12 As shown.

[0139] S403. Perform clustering on the sixth set of points and retain the top N clusters with the most nodes; where N = number of teeth - 1. In this embodiment, the DBSCAN clustering algorithm is used to accurately remove noise points and outliers. For example, in this embodiment, there are 14 teeth, so the top 13 clusters with the most nodes are retained.

[0140] S404. Using a plane fitting algorithm based on principal component analysis, plane fitting is performed on the mesh nodes within N clusters to obtain N cutting planes; for example... Figure 13 As shown, a total of 13 cutting planes are obtained in this embodiment.

[0141] S405. The crown region mesh model is divided using 13 cutting planes to obtain the following... Figure 14 The mesh model of the 14 crowns shown. Figure 14 The numbers 1-14 marked on each tooth crown represent the number of each tooth crown.

[0142] Finally, the boundaries of the mesh model for each crown were smoothed, resulting in the final crown mesh model as shown below. Figure 15 As shown.

[0143] It should be noted that the appendix to this application Figure 5 To be continued Figure 15 In this diagram, the unit for each coordinate axis is mm.

[0144] Example 3:

[0145] An automated full-mouth crown segmentation system based on gingival line recognition and region growth, used to implement the automated full-mouth crown segmentation method described in any of the above embodiments, the system comprising:

[0146] Preprocessing module: Used to preprocess the dental arch model obtained from oral scanning to obtain an initial mesh model;

[0147] Gingival line recognition module: used to identify the gingival line from the initial mesh model and obtain the mesh nodes of the gingival line region;

[0148] Crown extraction module: used to segment the gingival and crown regions based on the gingival line region mesh nodes, extract the crown, and obtain a crown region mesh model;

[0149] The crown segmentation module is used to segment the crown region mesh model into individual crowns, obtaining the mesh model of each individual crown.

[0150] The specific embodiments described above further illustrate the purpose, technical solution, 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. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

[0151] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

Claims

1. A method for automatic segmentation of full-mouth crowns based on gingival line recognition and regional growth, characterized in that, Includes the following steps: S1. Preprocess the dental arch model obtained from the oral cavity scan to obtain the initial mesh model; S2. Identify the gingival line based on the initial mesh model to obtain the mesh nodes of the gingival line region; S3. Based on the gingival line region mesh nodes, the gingival and crown regions are segmented to obtain the crown region mesh model; S4. Divide the tooth crown region mesh model to obtain the mesh model of each individual tooth crown.

2. The automatic segmentation method for full-mouth crowns based on gingival line recognition and region growth according to claim 1, characterized in that, Step S1 specifically includes: S101. Retain all the crowns and the gingiva connected to the crowns in the dental arch model obtained by oral scanning to obtain the first mesh model; S102. Perform coordinate registration on the first mesh model so that the gingival bottom surface of the first mesh model is coplanar with the XY plane, and the Z-axis extends upward and toward the direction of the crown, thus obtaining the second mesh model. S103. Perform adaptive smoothing on the second mesh model to obtain the initial mesh model.

3. The automatic segmentation method for full-mouth crowns based on gingival line recognition and region growth according to claim 1, characterized in that, Step S2 specifically includes: S201. Calculate the average curvature of all grid nodes in the initial grid model; S202. Select the grid nodes with the minimum average curvature in their neighborhood to obtain the first point set; S203. Select grid nodes with an average curvature less than the first threshold from the initial grid model to obtain the second point set; S204. Calculate the average angle of the normal vectors of all grid nodes in the neighborhood in the initial grid model; filter the grid nodes whose average angle of the normal vectors is greater than the second threshold to obtain the third point set; S205. Set the coordinate range of the gingival line in the Z-axis direction, filter the grid nodes located in the coordinate range, and obtain the fourth point set; S206. Take the intersection of the first point set, the second point set, the third point set, and the fourth point set to obtain the fifth point set; S207. Perform clustering processing on the fifth point set to remove outliers and noise points, and obtain the gingival line region mesh nodes.

4. The automatic segmentation method for full-mouth crowns based on gingival line recognition and region growth according to claim 3, characterized in that, In step S201, the average curvature of any grid node is calculated using the following formula: ； In the formula: H is the average curvature; E, F, and G are the first basic form coefficients corresponding to the grid node; L, M, and N are the second basic form coefficients corresponding to the grid node; In step S204, the average angle between the normal vectors of any grid node in its neighborhood is calculated using the following formula: ； In the formula: The average angle between the normal vectors of node i in its neighborhood; N(i) represents the set of neighboring nodes of node i; |N(i)| is the number of neighboring nodes; j represents the j-th neighboring node; The normal vector representing node i; This represents the normal vector of the j-th neighboring node.

5. The automatic segmentation method for full-mouth crowns based on gingival line recognition and region growth according to claim 1, characterized in that, Step S3 specifically includes: S301. Using the region growing algorithm, the grid node with the largest Z-axis coordinate is selected as the first seed point on the initial grid model; S302. Using the first seed point as a reference, determine the boundary zone between the gingiva and the crown based on the gingival line region mesh nodes; S303. Based on the grid nodes within the boundary band, the gingival and crown regions are segmented to obtain a grid model of the crown region.

6. The automatic segmentation method for full-mouth crowns based on gingival line recognition and region growth according to claim 5, characterized in that, Step S302 specifically includes: S3021. Starting from the first seed point, the grid grows outwards to the surrounding grid nodes, and all boundary nodes are determined using the following method: If the distance from a certain grid node to all grid nodes in the gingival region is greater than or equal to the third threshold, then the grid node is marked as a crown and continues to grow towards the surrounding grid nodes; If the distance from a certain grid node to any grid node in the gingival region is less than the third threshold, then the grid node is marked as a boundary node and growth stops; S3022. Starting from each boundary node, extend the grid to the unvisited grid nodes by a specified number of layers, and mark all grid nodes within the extension area as boundary nodes; S3023. The set of all grid nodes marked as boundary nodes is used as the boundary zone between the gingiva and the crown.

7. The automatic segmentation method for full-mouth crowns based on gingival line recognition and region growth according to claim 6, characterized in that, Step S303 specifically includes: S3031. Within the boundary zone between the gingiva and the crown, select the grid node with the largest Z-axis coordinate as the second seed point; S3032. Within the boundary zone between the gingiva and the crown, starting from the second seed point, the mesh nodes grow outwards to the surrounding mesh nodes, and the mesh nodes within the boundary zone are divided using the following method: If the average curvature of a certain grid node is greater than or equal to the fourth threshold, then the grid node is marked as a crown and continues to grow towards the surrounding grid nodes; If the average curvature of a grid node is less than the fourth threshold, the grid node is marked as non-crown and growth is stopped. S3033. Combine all the mesh nodes marked as crowns to obtain the crown region mesh model.

8. The automatic segmentation method for full-mouth crowns based on gingival line recognition and region growth according to claim 1, characterized in that, In step S4, the method for segmenting the tooth crown region mesh model includes: S401. Calculate the average angle between the normal vectors of all grid nodes in the neighborhood within the mesh model of the crown region; S402. Based on the average angle between the normal vectors of all grid nodes in their neighborhood, filter the grid nodes in the gap region to obtain the sixth point set. S403. Perform clustering on the sixth point set and retain the top N clusters with the most nodes; where N = number of teeth - 1; S404. Perform plane fitting on the mesh nodes in each of the N clusters to obtain N cutting planes; S405. The tooth crown region mesh model is divided by N cutting planes to obtain (N+1) tooth crown mesh models, and the boundary of the mesh model of each tooth crown is smoothed.

9. The automatic segmentation method for full-mouth crowns based on gingival line recognition and region growth according to claim 8, characterized in that, In step S402, the method for filtering grid nodes in the interdental region is as follows: if the average angle of the normal vectors of a certain grid node in its neighborhood is greater than the fifth threshold, then it is marked as a grid node in the interdental region.

10. The automatic segmentation method for full-mouth crowns based on gingival line recognition and region growth according to claim 8, characterized in that, The clustering process in step S403 uses the DBSCAN clustering algorithm; the plane fitting in step S404 uses a plane fitting algorithm based on principal component analysis.

Images