Method for tooth model segmentation and orthodontic scheme design, orthodontic appliance, and manufacture method

Abstract

Embodiments of the present invention relate to the technical field of digital orthodontics. Disclosed are a method for tooth model segmentation and orthodontic scheme design, an orthodontic appliance, and a manufacture method. In the present invention, gingivae and crowns are separated from a tooth model to be segmented to obtain gingiva data and crown data from said tooth model; starting from the innermost side of a dentition, the crowns of the innermost teeth are sequentially separated from the current crown data to obtain a crown separation result; a segmentation result of said tooth model is obtained on the basis of the crown separation result and the gingiva data. On the basis of the relatively significant shape difference between the innermost tooth in the dentition and a combination of all the other teeth, the present invention uses deep learning to recursively segment the tooth model, starting from the innermost side, so that the segmentation of the teeth does not depend on the appearance and shape of the teeth, thereby greatly improving the accuracy of the segmentation of the three-dimensional tooth model.

Description

Dental model segmentation and treatment plan design method, aligner and manufacturing method Cross-reference to Related Applications

[0001] This application is based on the Chinese patent application with the application number "2024110620517", the application date of August 2, 2024, and claims the priority of the above-mentioned Chinese patent application. The entire content of the above-mentioned Chinese patent application is incorporated herein by reference. TECHNICAL FIELD

[0002] Embodiments of the present application relate to the field of digital orthodontic technology, in particular to a dental model segmentation and treatment plan design method, an aligner and a manufacturing method. BACKGROUND

[0003] Accurate dental segmentation on digital three-dimensional (3D) dental models is a fundamental task in computer-aided orthodontic treatment. Accurate segmentation information can help dentists make clinical diagnoses, provide digital tooth shape information for patient orthodontic surgery, quantify the difference between expected and clinical treatment results, etc. However, accurately segmenting each tooth from the gingiva is a challenging problem, which is mainly reflected in the following aspects: (1) Although most people's teeth have some common geometric features, their shapes are unique and vary greatly between individuals; (2) The teeth of orthodontic patients are usually special, and their teeth often have missing, crowded or misaligned conditions, which can produce ambiguous tooth boundaries; (3) Noise in the scanning process can cause partial loss of tooth model surface.

[0004] To cope with the above challenges, traditional methods usually segment the dental model by using pre-selected geometric properties, such as coordinates, normal vectors or curvatures. However, the effectiveness of these traditional methods is more dependent on professional knowledge and experience. At the same time, the robustness of these methods is poor because using only those low-level geometric properties cannot accurately segment teeth with extreme appearances. Based on this, deep learning-based methods are proposed to learn task-oriented feature representations to achieve automatic segmentation of dental models.

[0005] The inventors find that at least the following problems exist in the prior art: since incising teeth from a three-dimensional tooth model is a tedious step, conventional deep learning schemes are basically focused on tooth segmentation in an end-to-end manner according to the appearance of the shape of the tooth, which is not only limited by the limitations of the model structure and the limitations of the data labeling by the labeling personnel, but also the prediction result is easily confused due to the similar appearance of the shape of some adjacent teeth, and the effect of using a conventional deep learning scheme to incise teeth from a three-dimensional tooth model is shown in FIG. 1, it can be clearly seen that in the position of the No. 4 tooth on the right, part of the surface patch that should belong to the No. 4 tooth is identified as a triangular surface patch of the No. 5 tooth, that is, the triangular surface patch of the No. 5 tooth is represented by dark green in FIG. 1, but there is also a dark green triangular surface patch on the No. 4 tooth, which is sufficient to show that the accuracy of the conventional deep learning scheme for segmenting a three-dimensional tooth model is often not high. SUMMARY

[0006] The purpose of the embodiments of the present application is to provide a tooth model segmentation method, an orthodontic scheme design method, an orthodontic device and a manufacturing method, which greatly improves the accuracy of segmenting a three-dimensional tooth model by deep learning to segment a tooth model from the innermost side in multiple stages.

[0007] To solve the above technical problems, the present application provides a tooth model segmentation method, comprising: segmenting the gums and crowns in the tooth model to be segmented to obtain gum data and crown data of the tooth model to be segmented; starting from the innermost side of the dentition, sequentially segmenting the innermost teeth in the current crown data to obtain a crown segmentation result; and combining the crown segmentation result and the gum data to obtain a segmentation result of the tooth model to be segmented. The present application considers that the shape of the innermost tooth in the dentition is significantly different from the combination of all teeth except the innermost tooth, and recursively segments the tooth model from the innermost side by deep learning, so that the segmentation of the tooth is not dependent on the appearance of the shape of the tooth, and the phenomenon of incorrect identification of the triangular surface patch of the adjacent teeth does not occur, thereby greatly improving the accuracy of segmenting a three-dimensional tooth model.

[0008] In some embodiments, the step of starting from the innermost side of the dentition, sequentially segmenting the innermost teeth in the current crown data to obtain a crown segmentation result comprises: sequentially recursively segmenting at least one innermost tooth in the current crown data by using a trained crown recursive tooth incision model, the crown recursive tooth incision model being a three-classification or two-classification neural network model, and the present embodiment can improve the efficiency of segmentation by using a trained crown recursive tooth incision model to recursively segment the innermost tooth.

[0009] In some embodiments, the crown recursive tooth segmentation model is trained by using crown data of a plurality of historical patients as training samples, wherein the innermost tooth crown of each training sample on any side is labeled as two classification categories respectively for training. The embodiment uses a binary classification neural network model to segment one innermost tooth at a time, with high segmentation accuracy.

[0010] In some embodiments, the crown recursive tooth segmentation model is trained by using crown data of a plurality of historical patients as training samples, wherein the innermost tooth crown of each training sample on any side is labeled as two classification categories respectively for training. The embodiment uses a binary classification neural network model to segment one innermost tooth at a time, with high segmentation accuracy.

[0011] In some embodiments, each tooth crown in the crown segmentation result is labeled by artificial means or by using a tooth automatic labeling algorithm, and the corresponding identification information is configured for each tooth crown. The embodiment labels the segmented tooth model to facilitate subsequent design of the orthodontic treatment plan.

[0012] In some embodiments, the tooth automatic labeling algorithm is used to label and configure the corresponding identification information for each tooth crown, including: inputting the crown data into the trained overall crown segmentation model to obtain a crown overall segmentation result, the crown overall segmentation result including each tooth crown after segmentation of the crown data and the corresponding identification information of each tooth crown; and matching one-to-one identification information for each tooth crown in the crown segmentation result through the crown overall segmentation result. The embodiment matches the segmentation result of the crown data by using the crown segmentation and labeling result of the overall crown segmentation model to achieve labeling, thereby achieving the purpose of automatic labeling.

[0013] In some embodiments, the matching one-to-one identification information for each tooth crown in the crown segmentation result through the crown overall segmentation result includes: calculating the probability of each tooth crown in the crown segmentation result corresponding to each crown category in the crown overall segmentation result; and determining the corresponding identification information of each tooth crown in the crown segmentation result based on the calculation result.

[0014] Optionally, the calculating the probability of each tooth crown in the crown segmentation result corresponding to each crown category in the overall crown segmentation result comprises: determining the probability of a current tooth crown corresponding to each crown category in the overall crown segmentation result according to the index coordinates of the triangular facets of the tooth crown in the crown segmentation result, and forming a probability matrix of each tooth crown in the crown segmentation result corresponding to each crown category in the overall crown segmentation result as a calculation result; and the determining the identification information corresponding to each tooth crown in the crown segmentation result based on the calculation result comprises: obtaining the identification information corresponding to each tooth crown in the crown segmentation result based on the probability matrix.

[0015] Optionally, the obtaining the identification information corresponding to each tooth crown in the crown segmentation result based on the probability matrix comprises: constantly selecting the largest element value in the probability matrix and deleting the row and column corresponding to the largest element value in the probability matrix after the selection; and determining the identification information corresponding to each tooth crown in the crown segmentation result according to the row number and column number of the largest element value selected in the probability matrix. In this embodiment, the label of each tooth crown is determined by calculating the probability of each tooth crown triangular facet in the segmentation result of the crown data corresponding to each crown category in the overall crown segmentation result, and the label result is objective and accurate.

[0016] In some embodiments, the configuring the corresponding identification information for each tooth crown based on the label result of each tooth crown further comprises: calculating the centroid coordinates of each tooth crown of the standard dentition based on the crown data of the standard dentition, obtaining a standard matching template of each tooth crown based on the centroid coordinates, calculating the centroid coordinates of each tooth crown data in the crown segmentation result, and obtaining the label result of each tooth crown based on the centroid coordinates of each tooth crown data and the corresponding standard matching template, and configuring the corresponding identification information for each tooth crown based on the label result of each tooth crown. In this embodiment, the standard matching template of each tooth crown is formed by using the standard dentition, and the determination of the label of each tooth crown is realized by using the standard matching template, which is simple and efficient with small calculation amount.

[0017] Optionally, the method further comprises: calculating a centroid coordinate of each dental crown data of the dental crown segmentation result, and converting the centroid coordinate into a unit vector; inputting the unit vector of each dental crown data into a trained machine learning model to obtain a label result of each dental crown, and configuring corresponding identification information for each dental crown based on the label result of each dental crown.

[0018] In some embodiments, the method further comprises:

[0019] Before combining the dental crown segmentation result and the gingival data, the segmentation data corresponding to each dental crown in the dental crown segmentation result is corrected by running a trained binary classification wisdom tooth molar adhesion model.

[0020] The wisdom tooth molar adhesion model is used to segment and delete the adhesion wisdom tooth part in the segmentation data corresponding to each dental crown in the dental crown segmentation result.

[0021] The embodiment optimizes the dental crown segmentation result by the wisdom tooth molar adhesion model, removes the possible adhesion wisdom tooth in the dental crown segmentation result, and makes the dental crown segmentation result more accurate.

[0022] In some embodiments, the gingival and dental crown in the to-be-segmented dental model are segmented to obtain the gingival data and the dental crown data of the to-be-segmented dental model, including:

[0023] The gingival and dental crown in the to-be-segmented dental model are segmented by a trained gingival and dental crown segmentation model to obtain preliminary segmentation data.

[0024] The preliminary segmentation data is post-processed by a graph cut method to correct the contour line in the preliminary segmentation data, and the gingival data and the dental crown data of the to-be-segmented dental model are obtained.

[0025] The trained gingival and dental crown segmentation model is obtained by labeling and training the gingival and dental crown in the historical user dental model.

[0026] After the gum and crown are segmented by using the gum crown segmentation model, the segmentation result is optimized by using the graph cut method, so that the segmentation result of the crown and gum is more accurate.

[0027] The application further provides an electronic device, comprising: at least one processor; and a memory in communication connection with the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the tooth model segmentation method.

[0028] The application further provides a computer readable storage medium storing a computer program, and the computer program is executed by a processor to implement the tooth model segmentation method.

[0029] The application further provides a treatment plan design method, comprising: establishing a three-dimensional tooth model according to the tooth model establishment method; and designing a treatment plan by taking the three-dimensional tooth model as an initial tooth model to obtain a series of target dental arch digital models for gradually moving a patient's dentition from a first tooth layout to a second tooth layout.

[0030] The application further provides a treatment device manufacturing method, comprising: manufacturing corresponding dental treatment devices according to the series of target dental arch digital models designed by the treatment plan design method.

[0031] The application further provides a treatment device, comprising: a dental treatment device manufactured by the treatment device manufacturing method. BRIEF DESCRIPTION OF DRAWINGS

[0032] One or more embodiments are illustrated by way of example in the drawings that are not intended to be limiting of the application, and the same reference numbers in different drawings denote the same elements, and the drawings are not necessarily to scale, unless otherwise specified. The drawings are intended to facilitate understanding of examples of implementations of the application that result in one or more advantages.

[0033] FIG. 1 is a segmentation effect diagram of tooth segmentation in an end-to-end manner according to the related art;

[0034] FIG. 2 is a flowchart of a tooth model segmentation method according to an embodiment of the application;

[0035] FIG. 3 is a diagram of a data structure of segmented crowns according to an embodiment of the application;

[0036] FIG. 4 is a diagram of a data structure of segmented gums according to an embodiment of the application;

[0037] FIG. 5 is a diagram of a neural network structure according to an embodiment of the application;

[0038] FIG. 6 is a schematic diagram of a training sample structure of a tooth crown recursive segmentation model according to an embodiment of the present application;

[0039] FIG. 7 is a schematic diagram of a training sample structure of a tooth crown recursive segmentation model according to another embodiment of the present application;

[0040] FIG. 8 is a schematic diagram of a tooth crown segmentation result according to an embodiment of the present application;

[0041] FIG. 9 is a schematic diagram of a training sample of a first kind of wisdom tooth molar adhesion model according to an embodiment of the present application;

[0042] FIG. 10 is a schematic diagram of a training sample of a second kind of wisdom tooth molar adhesion model according to an embodiment of the present application;

[0043] FIG. 11 is a schematic diagram of an optimization effect of a wisdom tooth molar adhesion model according to an embodiment of the present application;

[0044] FIG. 12 is a schematic diagram of a splicing result of a tooth crown segmentation result and gum data according to an embodiment of the present application;

[0045] FIG. 13 is a schematic diagram of a structure of an electronic device according to another embodiment of the present application. DETAILED DESCRIPTION

[0046] In order to make the objectives, technical solutions and advantages of the embodiments of the present application clearer, the embodiments of the present application will be described in detail below with reference to the drawings. However, those skilled in the art can understand that, in the embodiments of the present application, many technical details are presented in order to make the readers better understand the present application. However, the technical solutions claimed by the present application can be implemented even without these technical details and based on various changes and modifications of the following embodiments. The division of the following embodiments is for the convenience of description, and should not constitute any limitation on the specific implementation of the present application, and the embodiments can be combined and referenced with each other on the premise of no contradiction.

[0047] An embodiment of the present application relates to a tooth model segmentation method, which can be applied to any electronic device capable of segmenting tooth models, such as mobile phones, computer devices, and special instruments for collecting dental data. In this embodiment, the gingiva and crown in the tooth model to be segmented are segmented to obtain the gingiva data and crown data of the tooth model to be segmented; the innermost teeth in the current crown data are sequentially segmented from the innermost side of the dentition to obtain the crown segmentation result; and the segmentation result of the tooth model to be segmented is obtained by combining the crown segmentation result and the gingiva data. Since the innermost teeth in the dentition have a large shape difference relative to the teeth other than the two innermost teeth, the tooth model is segmented in multiple stages from the innermost side through deep learning, so that the segmentation of the teeth is not dependent on the appearance shape of the teeth, and the phenomenon of incorrect identification of triangular patches of adjacent teeth does not occur, thereby greatly improving the accuracy of the segmentation of the three-dimensional tooth model. The implementation details of the tooth model segmentation method of this embodiment are described below, and the following content is only provided for the implementation details for easy understanding, and is not necessary for implementing the present application.

[0048] As shown in FIG. 2, in step 101, the computer device for segmenting the tooth model segments the gingiva and crown in the tooth model to be segmented to obtain the gingiva data and crown data of the tooth model to be segmented; in some examples, the segmented crown data and gingiva data are shown in FIGS. 3 and 4, respectively.

[0049] In some examples, the process of segmenting the gingiva and crown in the tooth model to be segmented to obtain the gingiva data and crown data of the tooth model to be segmented can be: segmenting the gingiva and crown in the tooth model to be segmented through a trained gingiva and crown segmentation model to obtain preliminary segmentation data; and post-processing the preliminary segmentation data through a graph cut method to correct the contour lines in the preliminary segmentation data to obtain the gingiva data and crown data of the tooth model to be segmented; wherein the trained gingiva and crown segmentation model is obtained by labeling and training the gingiva and crown in the historical user tooth model.

[0050] In this embodiment, the graph cut method is used for the optimization and adjustment of the contour lines after the segmentation of the teeth and gingiva, and the main principle is that the graph cut method regards the image as a graph model, measures the segmentation result by using a penalty loss function for measuring the segmentation quality by giving prior hints, and in this example, the segmentation of the gingiva and crown in the tooth model to be segmented uses the two-class probability matrix after the two-classification of the gingiva and crown as the prior information of the graph cut method, so as to optimize the segmentation result of the crown and gingiva.

[0051] In step 102, the computer device for segmenting the tooth model sequentially segments the innermost tooth crown in the current crown data from the innermost side of the dentition to obtain a crown segmentation result; in oral medicine, the roots of the upper and lower jaw teeth grow in the alveolar fossa, and the crowns of the teeth abut each other in a certain order, direction and position, and when the segmentation does not start, the innermost tooth of the dentition is the tooth at the end of the dentition (usually the wisdom tooth in the case of no missing teeth), and the outermost tooth of the dentition is the tooth closest to the front of the oral cavity (lip) (usually the incisor tooth in the case of no missing teeth). Since the segmentation is sequentially performed, the innermost tooth will also change in a certain order, but it is always the tooth at the end of the dentition among all the teeth.

[0052] The present inventors found that the shape of the innermost tooth of each dentition is quite different from the overall shape formed by all the teeth except the two innermost teeth, and its spatial coordinates are relatively easy to segment. In view of this, the tooth model is recursively segmented from the innermost side through deep learning, which can make the segmentation of the tooth independent of the appearance shape of the tooth, and will not cause the phenomenon of incorrect identification of triangular patches of adjacent teeth. In some examples, the generated crown segmentation result is shown in FIG. 8, which clearly shows the crowns of different teeth after segmentation.

[0053] In some examples, the crowns of the innermost teeth in the crown data are segmented sequentially from the innermost to the outermost teeth in the dentition. This crown segmentation can be achieved by using a trained recursive incisor model to recursively segment at least one innermost tooth crown in the current crown data. This recursive incisor model is a three-class or two-class neural network model. The neural network model used is the Through Structure Global Correlation Network (TSGCNet) as an example. The neural network structure is shown in Figure 5. The network input data is a triangular mesh in mesh format. The input features are mainly divided into two parts: one part consists of the position coordinates of the vertices and center of the triangles, called coordinates, which flow through the C-stream channel in the figure; the other part consists of the vectors of the vertex normals and face normals of the triangles, called normal vectors, which flow through the N-stream channel in the figure. M represents the total number of input facets, and the dimension of the input features for each channel is 12. There are C-Stream and N-Stream channels. The C-Stream channel extracts features from the input through three attention-based layers (Attention-based Layers 1 / 2 / 3), while the N-Stream channel extracts features from the input through three max-pooling layers (Max-pooling-based Layers 1 / 2 / 3). The output of each layer... The dimensions are successively increased to 64, 128, and 256. After fusing the attention mechanism layers (increasing the dimension to 448), a 512-dimensional feature is obtained through a fully connected MLP (Multilayer Perceptron). Then, the features from the C-Stream and N-Stream channels are fused, and the fused features are further processed through a fully connected MLP to obtain the final segmentation result. For the C-Stream channel, the attention mechanism layers include KNN graph (k-nearest neighbor algorithm graph) and Graph attention. Specifically, KNN graph refers to constructing a graph using the center points of triangular facets and then selecting the k nearest nodes as connection points for the model network. Graph attention refers to assigning a weight parameter w to each network node. i (w1~w 6), and then the weight parameters of each node are used to weight average the corresponding node features, the blue dot represents a neighbor mesh cell, and the red dot represents a central mesh cell. Each attention mechanism layer of the N-Stream channel includes a KNN graph (k-nearest neighbor algorithm graph) and a Graph MaxPolling (graph maximum pooling layer), which refers to taking the weight maximum in different network node parameters, directly taking the maximum feature value of the weight, in the figure, the red arrow represents graph construction, the blue arrow represents skip connection, and the green arrow represents shared graph, represents concatenation.

[0054] When the tooth crown recursive incisor model is a three-class neural network model, the innermost teeth on both sides of the tooth crown data are directly segmented out, and when the tooth crown recursive incisor model is a two-class neural network model, each tooth crown in the current tooth crown data is recursively segmented out one by one from the innermost side.

[0055] In some examples, the crown recursive incisor model is trained by using crown data of a plurality of historical patients as training samples, wherein each training sample is trained by labeling the innermost teeth on both sides and all the teeth between the innermost teeth as three classification categories. The three classification categories are used for classification training, which can facilitate subsequent analysis and troubleshooting and reduce manual post-processing of incisors. Specifically, as shown in FIG. 6, in one segmentation training, the innermost teeth on both sides of each dental arch in the training sample are labeled as the 0th category and the 2nd category (FIG. 6 shows that there is a clear difference between the innermost two teeth and other teeth), and the overall shape of the other teeth is labeled as the 1st category. Then, the segmented innermost teeth (0th category and 2nd category) are deleted to achieve segmentation of the innermost teeth. In some examples, in addition to using a large amount of historical patient crown data as training samples for training, for cases where the historical patient crown data is less, the overall shape of the other teeth labeled as the 1st category in the previous segmentation can be used as input after one segmentation of the crown data of a historical patient, and the above labeling and segmentation operations can be repeatedly performed. In this way, the crown data of each historical patient can provide a plurality of different dental arch data for training as training samples. The crown recursive segmentation model is trained according to the dental arch data for training. Similarly, in other examples, the crown recursive incisor model is trained by using crown data of a plurality of historical patients as training samples, wherein each training sample is trained by labeling the innermost tooth on one side and all the other teeth as two classification categories. Specifically, in one segmentation, the innermost tooth on one side (e.g., the right side) of each dental arch in the training sample is labeled as the 0th category, and the other teeth are labeled as the 1st category (FIG. 7 shows that there is a clear difference between the innermost tooth on the right side of the dental arch and the other teeth), and after the innermost tooth is identified and segmented, the innermost tooth (0th category) is deleted to achieve segmentation of the innermost tooth. Since there is a large difference between the 0th category and the 1st category, a more accurate incisor effect can be obtained, and this operation can also be repeatedly performed so that one dental arch can produce a plurality of data, and then the data set is used to train the crown recursive segmentation model.

[0056] The above crown segmentation result only segments out individual crowns, but the purpose of segmenting the crowns is not only to segment the individual crowns, but more importantly, to design a subsequent treatment plan. For the design of the subsequent treatment plan, each crown needs to be assigned corresponding identification information for differentiation. The most commonly used identification information is the tooth number determined according to the FDI (Fédération Dentaire Internationale, International Dental Federation) tooth position notation. Therefore, in some embodiments, each tooth crown in the crown segmentation result is labeled by manual labeling or using a tooth automatic labeling algorithm, and each tooth crown is assigned corresponding identification information. The identification information refers to the classification of teeth according to the tooth morphology, and then an objective label is given to facilitate subsequent processing of the crown model, for example, labeling each crown in the crown segmentation result with an FDI number based on the FDI tooth position notation. In some embodiments, after obtaining the crown segmentation result, the crown segmentation result can be labeled directly by manual labeling, i.e., a designer labels each crown in the crown segmentation result with an FDI number according to the FDI tooth position notation. Manual labeling is simple and safe, and reduces further data processing of the incisors.

[0057] In other embodiments, the labeling process can also be automatically implemented using a tooth automatic labeling algorithm to further reduce the workload of the labeling personnel. Among them, the implementation of the tooth automatic labeling algorithm to assign corresponding identification information to each tooth crown has various implementation methods, and three implementation methods are given below:

[0058] The first method is to implement labeling based on the overall segmentation model. Specifically, in some examples, the crown data is input into the trained overall crown segmentation model to obtain the overall crown segmentation result, which includes each tooth crown segmented from the crown data and the corresponding identification information of each tooth crown. The overall crown segmentation result matches each crown in the crown segmentation result with one-to-one identification information.

[0059] The whole tooth crown segmentation model refers to a model that can directly predict the segmentation of all tooth crowns of FDI numbers at one time. The whole tooth crown segmentation model can be trained using tooth crown data of multiple historical patients as training samples. In some examples, the whole tooth crown segmentation model can be a 16-class neural network. In the training stage, the tooth crown data of the training samples is input into the 16-class neural network after being labeled respectively. After a large number of training samples are trained, the whole tooth crown segmentation model can recognize and segment the tooth crown model to be segmented. The segmentation result can not only segment a single tooth crown, but also correspond to the identification information of each tooth crown. That is, the whole tooth crown segmentation model can complete the classification and labeling of each tooth crown at one time. Since the tooth crown data is directly segmented into each tooth crown at one time, the tooth crown segmentation result obtained by completing the classification and labeling of each tooth crown at one time is referred to as a whole tooth crown segmentation result in this paper. The whole tooth crown segmentation result output by the whole tooth crown segmentation model includes each tooth crown after segmentation of the tooth crown data and the identification information corresponding to each tooth crown.

[0060] In some examples, the one-to-one matching of the identification information of each tooth crown in the tooth crown segmentation result with the whole tooth crown segmentation result can be: calculating the probability of each tooth crown in the tooth crown segmentation result corresponding to each tooth crown category in the whole tooth crown segmentation result; and determining the identification information corresponding to each tooth crown in the tooth crown segmentation result based on the calculation result.

[0061] In some examples, for each tooth crown in the tooth crown segmentation result, the probability of the tooth crown corresponding to each tooth crown category in the whole tooth crown segmentation result can be: determining the probability of the current tooth crown corresponding to each tooth crown category in the whole tooth crown segmentation result according to the index coordinates of the triangular facets of each tooth crown in the tooth crown segmentation result, forming a probability matrix of each tooth crown in the tooth crown segmentation result corresponding to each tooth crown category in the whole tooth crown segmentation result as a calculation result; and the corresponding determination of the identification information corresponding to each tooth crown in the tooth crown segmentation result based on the calculation result can be: obtaining the identification information corresponding to each tooth crown in the tooth crown segmentation result based on the probability matrix.

[0062] Specifically, it is assumed that the label processing performed here uses the FDI tooth position recording method proposed by the Fédération Dentaire Internationale (FDI), that is, each tooth is represented by 2 Arabic numerals, the ten's digit represents the quadrant where the tooth is located, the right upper, left upper, left lower, and right lower four quadrants of the patient are rotated clockwise, and the permanent teeth are 1, 2, 3, and 4, and the deciduous teeth are 5, 6, 7, and 8; the unit's digit represents the position of the tooth: 1-8 represent the central incisor to the third molar of the permanent tooth in turn, and 1-5 represent the central incisor to the second molar of the deciduous tooth in turn. Then, the probability that the current tooth crown corresponds to each crown category in the overall crown segmentation result is determined according to the index coordinates of the triangular surface patches of each tooth crown in the crown segmentation result, that is, a total of 16 arrays are generated for each crown. There are only 16 labels for teeth in the FDI tooth position recording method, and each number in the array corresponds to the probability of the tooth being each label in the FDI tooth position recording method. For N teeth, repeat the operation to obtain an N*16 probability matrix. The row represents the tooth number, the column represents the FDI label, and the element in the matrix represents the probability corresponding to the tooth number and the FDI label.

[0063] In some examples, based on the probability matrix, the identification corresponding to each tooth crown in the crown segmentation result can be: constantly selecting the maximum element value in the probability matrix and deleting the row and column corresponding to the maximum element value in the probability matrix after selection; and determining the identification information corresponding to each tooth crown in the crown segmentation result according to the row number and column number of the maximum element value selected in the probability matrix. For example, it is assumed that a probability matrix formed by three teeth is as follows:

[0064] The maximum element value in the probability matrix is selected as 0.9, the row and column corresponding to the value are the third row and the third column, that is, the FDI number of the third tooth is 3, and the third row and the third column are deleted, and the matrix becomes:

[0065] The maximum value of the matrix is continuously searched as 0.8, that is, the FDI number of the second tooth is 1, and the second row and the first column are deleted, and the matrix becomes: 0 0 0 0 0 0 0 0 0 0 0 0.2 0.7 0.1

[0066] The maximum value is searched as 0.7, and the FDI number of the first tooth is 15. It is worth noting that 15 in this example refers to the 15th of the original 16-tooth matrix, not the sequence number 15 in this 1*14 matrix. This example only takes three teeth as an example, and the selection method for a probability matrix formed by multiple teeth is similar to this example, as long as the row is less than or equal to the column.

[0067] The second implementation mode of automatically labeling each tooth crown with corresponding identification information by using the tooth automatic labeling algorithm is a template matching mode, for example:

[0068] First, the centroid coordinate data of each tooth crown in a standard dentition (assuming that the standard dentition includes 16 teeth) is calculated according to all the tooth crown data of the standard dentition. The centroid coordinate of each tooth can be the average of the triangular facet coordinates of all teeth. The corresponding unit vector is calculated for the coordinates of the 16 centroids, that is, the centroid coordinates are normalized. For example, the centroid coordinates are (x, y, z), which are vectors relative to the origin The normalized result is Thus, the corresponding unit vector is obtained. The unit vectors corresponding to the 16 centroid coordinates are used as the standard matching templates of each tooth crown. Then, the centroid coordinates of each category of tooth crown data in the tooth crown segmentation result are calculated and converted into unit vectors. The included angle between the 16 unit vectors of the standard matching templates is calculated, and the tooth is assigned to the FDI number corresponding to the minimum angle vector.

[0069] The third implementation mode of automatically labeling each tooth crown with corresponding identification information by using the tooth automatic labeling algorithm can also be realized based on a machine learning mode, for example: first, the centroid coordinate data of all training data is calculated in the data preparation stage, and the centroid coordinates are converted into unit vectors. The FDI number corresponding to the unit vector is saved. Then, enter the training model stage. A machine learning model (such as a vector machine model or a random forest model) is trained according to the input (unit vector) and output (FDI number category). Finally, the centroid coordinates of the tooth crown data in the tooth crown segmentation result are calculated and converted into unit vectors, which are input into the model trained in the training model stage to obtain the FDI labeling result.

[0070] In some cases, the wisdom tooth of the patient can not be fully developed and is adhered to the adjacent tooth No. 7, as shown in FIG. 8. In this case, the segmentation result of tooth No. 7 in the above crown segmentation result can have the wisdom tooth adhered to tooth No. 7. In this case, the segmentation result of tooth No. 7 needs to be further optimized to remove the adhered wisdom tooth. Therefore, in some examples, before combining the crown segmentation result and the gingival data, the segmentation data corresponding to each tooth crown in the crown segmentation result is corrected by running the trained wisdom tooth-molar adhesion model. The wisdom tooth-molar adhesion model is used to segment and remove the adhered wisdom tooth part in the segmentation data corresponding to each tooth crown in the crown segmentation result. By using the wisdom tooth-molar adhesion model, it is ensured that there is no adhered wisdom tooth part on each tooth crown segmented in the crown segmentation result, further improving the accuracy of segmentation. The wisdom tooth-molar adhesion model is trained by using a plurality of training samples of wisdom tooth-molar adhesion, as shown in FIGS. 9 and 10, and then inputting the segmented molar part and wisdom tooth part into a binary classification neural network for training. For example, the segmentation result of the wisdom tooth part with adhesion in FIG. 8 is optimized by using the wisdom tooth-molar adhesion model to obtain the molar data as shown in FIG. 11.

[0071] In step 103, the computer device for segmenting the tooth model combines the crown segmentation result and the gingival data to obtain the segmentation result of the tooth model to be segmented.

[0072] In some examples, combining the crown segmentation result and the gingival data to obtain the segmentation result of the tooth model to be segmented can be: reading the gingival mesh data and recording it as mesh1, and reading the crown mesh data and recording it as mesh2; v1, f1, and c1 are respectively recorded as the vertices, face coordinate index, and face color of mesh1; similarly, v2, f2, and c2 are respectively recorded as the vertices, face coordinate index, and face color of mesh2. Each value of the matrix of f2 is added to the number of vertices of mesh1, i.e., the width of v1, then v1 and v2 are merged and recorded as V, i.e., V is the vertex of the spliced model; c1 and c2 are merged and recorded as C, i.e., C is the color of the spliced model; f1 and f2 are merged and recorded as F, i.e., F is the face index of the spliced model. According to F, V, and C, the final color-spliced model can be generated. In some examples, the segmentation result of the tooth model to be segmented formed is as shown in FIG. 12. The merging operation can be row splicing.

[0073] In some examples, after obtaining the segmentation result of the tooth model to be segmented, i.e., after splicing the dental crown and the gingiva into a complete three-dimensional model file, an upsampling stage can be entered. The upsampling stage refers to taking the complete three-dimensional model file spliced by the dental crown and the gingiva as a result of inference, training a support vector machine model according to the triangle facet center coordinates and categories extracted from the segmentation result of the tooth model to be segmented, and using the model to predict the original oral scanning model to obtain a final segmentation result. The support vector machine model (Support Vector Machine, abbreviated as SVM) is a supervised learning model mainly used for classification and regression analysis. The support vector machine model is a general learning method based on statistical learning theory, which classifies data by finding a maximum margin hyperplane.

[0074] In the embodiment, the gingiva and the dental crown in the tooth model to be segmented are segmented to obtain gingiva data and dental crown data of the tooth model to be segmented; the innermost teeth in the current dental crown data are sequentially segmented from the innermost side of the dental arch to obtain a dental crown segmentation result; and the segmentation result of the tooth model to be segmented is obtained in combination with the dental crown segmentation result and the gingiva data. Since it is noted that the innermost teeth in the dental arch have a larger shape difference relative to the teeth other than the two innermost teeth, the tooth model is segmented from the innermost side in multiple stages through deep learning, so that the segmentation of the teeth is not dependent on the appearance shape of the teeth, and thus the accuracy of the segmentation of the three-dimensional tooth model is greatly improved.

[0075] The step division of the above method is only for the purpose of clear description, and in implementation, one step can be combined or some steps can be split and decomposed into multiple steps, as long as the same logical relationship is included, and all are within the protection scope of the present application; adding irrelevant modifications or introducing irrelevant designs in the algorithm or process, but not changing the core design of the algorithm and process, are within the protection scope of the present application.

[0076] Another embodiment of the present application relates to an electronic device, as shown in FIG. 13, comprising at least one processor 501; and a memory 502 in communication connection with the at least one processor; wherein the memory 502 stores instructions executable by the at least one processor 501, and the instructions are executed by the at least one processor 501 to enable the at least one processor 501 to perform the tooth model segmentation method as described above.

[0077] The memory 502 and the processor 501 are connected by a bus in a bus manner, the bus can include any number of interconnected buses and bridges, the bus connects one or more processors 501 and various circuits of the memory 502 together. The bus can also connect various other circuits such as peripheral devices, voltage stabilizers and power management circuits together, which are well known in the art, and therefore, further description is not made herein. The bus interface provides an interface between the bus and the transceiver. The transceiver can be one element or multiple elements such as multiple receivers and transmitters, which provide units for communicating with various other devices on the transmission medium. The data processed by the processor 501 is transmitted on the wireless medium through the antenna, further, the antenna also receives data and transmits the data to the processor 501.

[0078] The processor 501 is responsible for managing the bus and general processing, and can also provide various functions, including timing, peripheral interface, voltage regulation, power management and other control functions. The memory 502 can be used to store data used by the processor 501 in performing operations.

[0079] Another embodiment of the present application relates to a computer readable storage medium, which stores a computer program. The computer program is executed by a processor to implement the above-mentioned method embodiment.

[0080] That is, those skilled in the art can understand that all or part of the steps in the above-mentioned embodiment method can be completed by a program instructing related hardware, the program is stored in a storage medium, and includes a plurality of instructions for causing a device (which can be a single-chip microcomputer, a chip, etc.) or a processor to execute all or part of the steps of the method described in various embodiments of the present application. The foregoing storage medium includes a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk and various program code storage media.

[0081] Another embodiment of the present application relates to a treatment plan design method, which can be applied to any electronic device that can generate a treatment plan, such as a mobile phone, a computer and the like. In the present embodiment, the electronic device that generates the treatment plan establishes a three-dimensional tooth model by the tooth model establishment method of the above-mentioned embodiment; then, the three-dimensional tooth model is used as an initial tooth model to design a treatment plan, and a series of target dental digital models for gradually moving the patient's dentition from a first tooth layout to a second tooth layout are obtained.

[0082] In some examples, a series of target dental arch digital models for progressively moving the patient's dentition from a first dental arrangement to a second dental arrangement can be obtained by a state space search method, which transforms the problem solving process into a process of finding a path from an initial state to a target state. The state space search method can be represented by a triple: [S, O, G], where S is the set of initial states, O is the set of operators, and G is the set of target states. Applying an operator to a current state will cause a change in state, thus reaching a new state. The set of all states can be represented by a state graph. In the state graph, some states are target states, and if a path from the start state to a target state is found, the problem is solved. The solution is the series of operators used to reach the target state from the start state. If the path is the shortest, the path can be called the optimal solution. Common state space searches are depth-first and breadth-first. Breadth-first is to find from the initial state layer by layer downward until the target is found. Depth-first is to find one branch first in a certain order, and then find another branch. The breadth-first and depth-first searches mentioned above are blind searches. The state space search method also includes heuristic searches. The state space search method can be used to develop an orthodontic treatment plan, in which the initial state can be the patient's original dental state; the set of operators can include null operation, tooth extraction, interproximal enamel reduction, translation, rotation, etc.; and the set of target states is the set of dental states that meet the orthodontic requirements developed by the clinician. A series of target dental arch digital models for progressively moving the patient's dentition from a first dental arrangement to a second dental arrangement can be obtained based on this principle.

[0083] In some examples, a series of target dental arch digital models for progressively moving the patient's dentition from a first dental arrangement to a second dental arrangement can be obtained by an interpolation tooth arrangement method, which can be a method of generating a dental arch curve corresponding to a three-dimensional tooth model, and for each pair of reference points on the dental arch curve, using interpolation or other mathematical methods to obtain a corresponding secondary normal axis for each point; the tangent axis and the secondary normal axis of each point form a tangent plane of the local coordinate system of the dental arch curve, the perpendicular of the tangent plane is the primary normal axis, the tangent axis and the primary normal axis form the osculating plane, and the primary normal axis and the secondary normal axis form the normal plane; finally, a natural coordinate system of the dental arch curve is established, in which the tangent axis, the secondary normal axis, the primary normal axis, the osculating plane, the tangent plane, and the normal plane are different at different points on the three-dimensional dental arch curve. Subsequently, under the adjustment scheme of the physician, the dental arch curve is adjusted and corrected, and automatic tooth alignment is performed based on the adjusted dental arch curve and the coordinate system of the dental arch curve.

[0084] Another embodiment of the present application relates to a method for manufacturing an appliance, which can be applied to any electronic industrial equipment or system that can manufacture an appliance. In this embodiment, the electronic industrial equipment or system for manufacturing an appliance manufactures the corresponding dental appliance according to the series of target dental digital models designed by the appliance design method proposed in the above embodiment.

[0085] In some examples, the corresponding dental appliance is manufactured in two ways, each of which is exemplified by an implementation. The first way can be: manufacturing a physical model of the target dental digital model by using additive manufacturing or 3D printing technology; and manufacturing a shell-shaped body corresponding to the dental appliance by using a hot stamping film according to the physical model of the target dental digital model. The second way can be: generating a digital model of the shell-shaped body corresponding to the dental appliance according to the target dental digital model; and manufacturing the shell-shaped body corresponding to the dental appliance by using additive manufacturing or 3D printing technology according to the digital model of the shell-shaped body corresponding to the dental appliance.

[0086] Another embodiment of the present application relates to an appliance, which includes: a dental appliance manufactured by using the appliance manufacturing method in the above embodiment. That is, the shell-shaped body corresponding to the dental appliance manufactured in the above embodiment.

[0087] It can be found that the present embodiment is an appliance corresponding to the above method embodiment, and the present embodiment can be implemented in cooperation with the above method embodiment. The related technical details mentioned in the above dental model segmentation method and appliance manufacturing method embodiments are still valid in the present embodiment, and are not described herein again in order to reduce repetition. Correspondingly, the related technical details mentioned in the present embodiment can also be applied to the above method embodiments.

[0088] It is worth mentioning that each module involved in the above embodiments is a logical module, which can be a physical unit, a part of a physical unit, or a combination of multiple physical units in actual application. In addition, in order to highlight the innovative part of the present application, units that are not closely related to solving the technical problems proposed in the present application are not introduced in the present embodiment, but this does not mean that there are no other units in the present embodiment.

[0089] Those skilled in the art can understand that the above embodiments are specific examples for implementing the present application, and in actual application, various changes can be made in form and details without departing from the spirit and scope of the present application.

Claims

1. A tooth model segmentation method, comprising: segmenting gingiva and crowns in a tooth model to be segmented to obtain gingiva data and crown data of the tooth model to be segmented; from the innermost side of the dentition, sequentially segmenting the innermost tooth crown in the current crown data to obtain a crown segmentation result; combining the crown segmentation result and the gingiva data to obtain a segmentation result of the tooth model to be segmented.

2. The dental cast segmentation method of claim 1, wherein, The step of sequentially segmenting the innermost tooth crown in the current crown data to obtain a crown segmentation result comprises: recursively segmenting at least one innermost tooth crown in the current crown data using a trained crown recursive incisor model, which is a three-class or two-class neural network model.

3. The dental cast segmentation method of claim 2, wherein, The crown recursive incisor model is trained using crown data of a plurality of historical patients as training samples, wherein the innermost tooth crown on the left and right sides of each training sample and all tooth crowns between the innermost teeth are labeled as three classification categories for training.

4. The dental cast segmentation method of claim 2, wherein, The crown recursive incisor model is trained using crown data of a plurality of historical patients as training samples, wherein the innermost tooth crown on any side of each training sample and all other tooth crowns are labeled as two classification categories for training.

5. The dental cast segmentation method of claim 1, wherein, Each tooth crown in the crown segmentation result is labeled by manual labeling or using a tooth automatic labeling algorithm.

6. The dental cast segmentation method of claim 5, wherein, The step of labeling each tooth crown using a tooth automatic labeling algorithm comprises: inputting the crown data into a trained overall crown segmentation model to obtain a crown overall segmentation result, which includes each tooth crown segmented from the crown data and the corresponding identification information of each tooth crown; matching the identification information of each tooth crown in the crown segmentation result with the corresponding identification information in the crown overall segmentation result.

7. The dental cast segmentation method of claim 6, wherein, The step of matching the identification information of each tooth crown in the crown segmentation result with the corresponding identification information in the crown overall segmentation result comprises: calculating the probability of each tooth crown in the crown segmentation result corresponding to each tooth crown category in the crown overall segmentation result; determining the identification information of each tooth crown in the crown segmentation result based on the calculation result.

8. The dental cast segmentation method of claim 7, wherein, The step of calculating the probability of each tooth crown in the crown segmentation result corresponding to each tooth crown category in the crown overall segmentation result comprises: determining the probability of the current tooth crown corresponding to each tooth crown category in the crown overall segmentation result according to the index coordinates of the triangular facets of each tooth crown in the crown segmentation result, and forming a probability matrix of each tooth crown in the crown segmentation result corresponding to each tooth crown category in the crown overall segmentation result as a calculation result. The step of determining the identification information of each tooth crown in the crown segmentation result based on the calculation result comprises: obtaining the identification information of each tooth crown in the crown segmentation result based on the probability matrix.

9. The dental cast segmentation method of claim 8, wherein, The identification information corresponding to each tooth crown in the crown segmentation result is obtained based on the probability matrix, and the identification information corresponding to each tooth crown in the crown segmentation result is obtained based on the probability matrix, including: The maximum element value in the probability matrix is selected constantly, and the row and column corresponding to the maximum element value in the probability matrix are deleted after being selected; The identification information corresponding to each tooth crown in the crown segmentation result is determined according to the row number and column number of the selected maximum element value in the probability matrix.

10. The dental cast segmentation method of claim 5, wherein, The tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling algorithm is used for labeling processing, and the tooth automatic labeling ​ ​ ​ ​ 11. The dental cast segmentation method of claim 5, wherein, ​ ​ ​ ​ ​ ​ ​ ​ 12. The dental cast segmentation method of any one of claims 1-11, wherein, ​ ​ ​ 13. The dental cast segmentation method of any one of claims 1-11, wherein, ​ ​ ​ ​ ​ ​ ​ ​ The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the tooth model segmentation method according to any one of claims 1 to 13.

15. A computer readable storage medium storing a computer program, the computer program being executed by a processor to implement the tooth model segmentation method according to any one of claims 1 to 13.

16. A method for designing an orthodontic treatment plan, comprising: establishing a three-dimensional tooth model according to the tooth model establishment method according to any one of claims 1 to 13; designing an orthodontic treatment plan with the three-dimensional tooth model as an initial tooth model to obtain a series of target dental arch digital models for gradually moving a patient's dentition from a first tooth arrangement to a second tooth arrangement.

17. A method for manufacturing an orthodontic appliance, comprising: manufacturing corresponding dental appliances according to the series of target dental arch digital models designed by the orthodontic treatment plan design method according to claim 16.

18. An appliance comprising: a dental appliance manufactured by the method for manufacturing an orthodontic appliance according to claim 17.

Images