Dental restoration generation method and device, computer equipment and storage medium

Abstract

The embodiment of the invention discloses a dental restoration generation method and device, computer equipment and a storage medium. The method comprises the following steps: acquiring a target tooth jaw model, a target tooth position number and a target prosthesis description text, wherein the target tooth position number is a tooth position number corresponding to a target tooth position of a to-be-designed dental prosthesis; and inputting the target tooth jaw model, the target tooth position number and the target restoration description text into a trained restoration generation model for feature extraction fusion processing, and generating a target dental restoration model. According to the embodiment of the invention, the dental restoration model is directly generated according to the text feature vector and the dental space feature vector of the user, the link of understanding and modeling depending on artificial experience is replaced, semantic understanding and output are directly and automatically performed, the design expectation of the user is deeply understood, the modeling process is simplified, and the modeling efficiency is improved. The technical problem that in the prior art, the design efficiency of the dental prosthesis is low is solved.

Description

Technical Field

[0001] This application relates to the field of dental computer-aided design, and in particular to methods, devices, computer equipment and storage media for the generation of dental prostheses. Background Technology

[0002] Teeth consist of a crown and a root. The crown is the part of the tooth that is visible in the oral cavity, while the root is the part embedded in the alveolar bone and not visible. Designing dental prostheses such as crowns and bridges heavily relies on the manual experience and skills of dental technicians. Technicians use computer-aided design (CAD) software to create long-term models based on plaster or digital models provided by the dentist. However, each dental prosthesis design is highly dependent on the dental technician's experience. Communication between dentists and technicians primarily involves written notes or verbal communication to convey design intentions. This subjective approach makes precise quantification difficult, leading to discrepancies between the final prosthesis design and expectations, resulting in repeated revisions that consume significant time and effort. While existing technologies utilize artificial intelligence (AI) in dental prosthesis design, they largely focus on automatically generating occlusal surfaces based on opposing teeth or performing simple morphological matching. They fail to understand and respond to the user's abstract and subjective stylistic needs, thus resulting in low efficiency in dental prosthesis design. Summary of the Invention

[0003] This application provides a method, apparatus, computer equipment, and storage medium for generating dental prostheses, which can solve the technical problem of low efficiency in dental prosthesis design.

[0004] In a first aspect, embodiments of this application provide a method for generating a dental prosthesis, comprising: Obtain the target dental model, target tooth position number, and target restoration description text, wherein the target tooth position number is the tooth position number corresponding to the target tooth position of the dental restoration to be designed; The target dental prosthesis model is generated by performing feature extraction and fusion processing on the target jaw model, the target tooth position number, and the target prosthesis description text as inputs to the trained prosthesis generation model.

[0005] In some embodiments, the step of performing feature extraction and fusion processing on the target dental prosthesis generation model after inputting the target jaw model, the target tooth position number, and the target prosthesis description text into the trained prosthesis generation model to generate the target dental prosthesis model includes: Input the target jaw model, the target tooth position number, and the target restoration description text into the restoration generation model; Based on the target dental model and the target tooth position number, feature extraction is performed to obtain the dental spatial feature vector, the tooth position number feature vector, and the tooth position spatial location feature vector. Feature extraction is performed on the description text of the target repair body to obtain the target text feature vector; The tooth position number feature vector, the tooth position spatial location feature vector, and the target text feature vector are subjected to feature fusion processing to generate the target dental restoration model.

[0006] In some embodiments, feature fusion processing is performed on the jaw space feature vector, tooth position number feature vector, tooth position spatial location feature vector, and target text feature vector to generate a target dental restoration model, including: The feature fusion processing is performed on the tooth position number feature vector, the tooth position spatial location feature vector, the spatial feature vector of the previously generated dental restoration, the jaw space feature vector, and the target text feature vector to obtain the spatial feature vector of the target dental restoration. Before feature fusion processing, it is determined whether the target restoration description text corresponding to the target text feature vector has been modified. If it has not been modified, the spatial feature vector of the previously generated dental restoration is set to 0. A model of the target dental restoration is generated based on the spatial feature vector of the target dental restoration.

[0007] In some embodiments, the restoration generation model includes a geometric encoder, a geometric decoder, a text encoder, and a conditional diffusion module.

[0008] In some embodiments, the step of performing feature fusion processing on the jaw space feature vector, the tooth position number feature vector, the tooth position spatial location feature vector, and the target text feature vector to generate a target dental restoration model includes: The spatial feature vector of the target dental restoration is obtained by performing feature fusion processing on the dental space feature vector, the tooth position number feature vector, the tooth position spatial location feature vector, and the target text feature vector through the conditional diffusion module and the geometric encoder. The geometric decoder generates a model of the target dental restoration based on the spatial feature vector of the target dental restoration.

[0009] In some embodiments, generating a target dental restoration model based on the spatial feature vector of the target dental restoration using the geometry decoder includes: The spatial feature vector of the target dental restoration is input into the geometric decoder for decoding to obtain an intermediate dental restoration model. When a user confirms the intermediate dental prosthesis model, or when no user modifies the intermediate dental prosthesis model is received within a preset time period, the intermediate dental prosthesis model is used as the target dental prosthesis model.

[0010] In some embodiments, generating a target dental restoration model based on the spatial feature vector of the target dental restoration using the geometry decoder includes: The spatial feature vector of the target dental restoration is input into the geometric decoder for decoding to obtain an intermediate dental restoration model. When a user receives a modification instruction for the first dental prosthesis model of the intermediate dental prosthesis model within a preset time period, the text encoder performs text feature extraction processing on the modification description text in the modification instruction for the first dental prosthesis model to obtain the modification text feature vector. The dental restoration is adjusted by the conditional diffusion module and the geometric encoder based on the modified text feature vector, the dentition space feature vector and the intermediate dental restoration space feature vector corresponding to the intermediate dental restoration model, to obtain the dental restoration adjustment space feature vector. A dental prosthesis adjustment model is generated based on the geometric decoder and the dental prosthesis adjustment space feature vector. When a user confirms the adjustment model of the dental prosthesis, or when no user modifies the adjustment model of the dental prosthesis is received within a preset time period, the adjustment model of the dental prosthesis is used as the target dental prosthesis model. When a user's second dental prosthesis model modification instruction is received within a preset time period, the dental prosthesis adjustment model is updated to an intermediate dental prosthesis model, the second dental prosthesis model modification instruction is updated to a first dental prosthesis model modification instruction, and the process of extracting text features from the modification description text in the first dental prosthesis model modification instruction using the text encoder to obtain the modification text feature vector is returned.

[0011] In some embodiments, the dentition space feature vector includes at least one of the following features: morphological features, size features, spatial position features, and occlusal relationship features of the associated teeth of the target tooth position. The associated teeth include the adjacent teeth, opposing teeth, and contralateral teeth of the target tooth position.

[0012] In some embodiments, the target dental model is obtained by: Obtained by scanning the oral cavity or plaster model of the target user using a target scanner, wherein the target scanner includes an oral scanner or a fixed scanner; Alternatively, the target dental model can be obtained through a local database or a cloud database.

[0013] In some embodiments, after generating the target dental restoration model based on the jaw space feature vector and the initial text feature vector, the method further includes: Send the target dental restoration model to a preset user terminal; Alternatively, the target dental prosthesis model can be sent to a 3D printing device for printing based on the selected material.

[0014] In some embodiments, the step of performing feature extraction and fusion processing on the target dental prosthesis generation model after inputting the target jaw model, the target tooth position number, and the target prosthesis description text into the trained prosthesis generation model to generate the target dental prosthesis model includes: The target dental prosthesis model, the target tooth position number, and the target prosthesis description text are input into the trained prosthesis generation model. Feature extraction and fusion processing are then performed to generate the target dental prosthesis model and / or the target dental prosthesis model rendering.

[0015] Secondly, embodiments of this application also provide a dental prosthesis generation device, which includes: The acquisition unit is used to acquire the target dental model, the target tooth position number, and the target restoration description text, wherein the target tooth position number is the tooth position number corresponding to the target tooth position of the dental restoration to be designed. The model generation unit is used to perform feature extraction and fusion processing on the target dental prosthesis model, the target tooth position number, and the target prosthesis description text input into the trained prosthesis generation model to generate the target dental prosthesis model.

[0016] Thirdly, embodiments of this application also provide a dental prosthesis generation computer device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the above-described method.

[0017] Fourthly, embodiments of this application also provide a computer-readable storage medium storing a computer program, the computer program including program instructions that, when executed by a processor, can implement the above-described method.

[0018] This application provides a method, apparatus, computer device, and storage medium for generating dental prostheses. The method includes: acquiring a target jaw model, a target tooth position number, and a target prosthesis description text, wherein the target tooth position number is the tooth position number corresponding to the target tooth position of the dental prosthesis to be designed; inputting the target jaw model, the target tooth position number, and the target prosthesis description text into a trained prosthesis generation model for feature extraction and fusion processing to generate a target dental prosthesis model. This application directly generates a target dental prosthesis model based on the user's description text and the target jaw model, replacing the step of relying on manual experience to understand and model, directly automating semantic understanding and output, deeply understanding the user's design expectations, simplifying the modeling process, and solving the technical problem of low efficiency in dental prosthesis design in the prior art. Attached Figure Description

[0019] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 A schematic flowchart illustrating the method for generating dental prostheses provided in this application embodiment; Figure 2 A schematic diagram of a sub-process of the dental prosthesis generation method provided in the embodiments of this application; Figure 3 This is another schematic diagram of a sub-process of the dental prosthesis generation method provided in the embodiments of this application; Figure 4 This is another schematic diagram of a sub-process of the dental prosthesis generation method provided in the embodiments of this application; Figure 5 This is another schematic diagram of a sub-process of the dental prosthesis generation method provided in the embodiments of this application; Figure 6 Another sub-process diagram of the dental prosthesis generation method provided in the embodiments of this application. Figure 7 A schematic block diagram of a dental prosthesis generation device provided in the embodiments of this application; Figure 8 A schematic block diagram of a computer device provided in an embodiment of this application. Detailed Implementation

[0021] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0022] It should be understood that, when used in this specification and the appended claims, the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0023] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the scope of the application. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0024] It should also be further understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0025] In the field of computer-aided dental design, patients request assistance from dentists in designing target restorations, which can be dental prostheses. Dental prostheses can be single-tooth restorations or combinations of multiple teeth. They include crowns (such as partial crowns, full crowns, single crowns, and multiple crowns), dentures, implants, bridges, and inlays. Dentists obtain target dental models by scanning the patient's oral cavity with a scanner or by surface scanning a plaster model of the mouth. Currently, dentists provide these target dental models to dental technicians, who then design the specific shape of the dental restoration using CAD software.

[0026] However, the experience of each dental technician varies, leading to fluctuations in the quality of dental restorations designed by different technicians. Typically, the initial design of a dental restoration by a technician is not immediately accepted by the client because design expectations, such as design style, are communicated verbally. The technician then proceeds with the design after understanding these expectations. In this manual communication process, it is difficult to guarantee a complete understanding of the client's design expectations, and thus, the accuracy of the design is uncertain. If the initial design does not meet the client's expectations, the technician continues to adjust it in CAD software based on the client's verbal descriptions. This back-and-forth design confirmation process is entirely manual, resulting in low design efficiency.

[0027] In addition, dental technicians must possess basic CAD skills, but these skills vary widely. Designing using CAD software is relatively slow, which in turn reduces the efficiency of designing dental restorations.

[0028] Existing technologies mostly focus on automatically generating occlusal surfaces based on opposing teeth or performing simple morphological matching. These methods cannot understand or respond to users' abstract and subjective stylistic needs, resulting in low design accuracy and consequently low design efficiency.

[0029] In summary, the current process for designing dental prostheses suffers from low design efficiency.

[0030] To address the aforementioned issues, embodiments of this application provide a method, apparatus, computer device, and storage medium for generating dental prostheses.

[0031] Figure 1 This is a schematic flowchart of the dental restoration fabrication method provided in the embodiments of this application. Figure 1 As shown, the method includes the following steps S110-S120.

[0032] S110. Obtain the target tooth jaw model, target tooth position number, and target restoration description text. The target tooth position number is the tooth position number corresponding to the target tooth position of the dental restoration to be designed.

[0033] In some embodiments, this dental prosthesis generation method is applied to a dental prosthesis generation computer device, which can be a terminal or a server. The terminal can be a smartphone, tablet, PDA, or laptop computer, etc. This dental prosthesis generation method is applicable to single-tooth, multi-tooth, or multi-type prosthesis combinations, such as bridges, inlays, single crowns, and multiple crowns. Target prostheses include crown-type prostheses, dentures, dental implants, etc., with crown-type prostheses including partial crowns and full crowns.

[0034] In some embodiments, the target repair body description text can be obtained in the following ways: When a user clicks on a dental prosthesis generation computer device, the device generates a dialog box prompting the user to input text describing the desired style of the dental prosthesis. The user then enters a descriptive text for the target prosthesis based on this dialog box. Alternatively, the user can communicate with the device via voice, verbally expressing their desired style. The device receives the user's voice input, uses a semantic understanding module to interpret the semantics, and converts it into a descriptive text for the target prosthesis. This converted descriptive text can be presented on the interactive interface in text form or directly as voice.

[0035] The system obtains the user's desired dental restoration style description text through either the dialog box or by converting speech to text. This triggers the generation of expected instructions containing the user's design preferences. Upon recognition of the expected instructions, the system parses them and begins designing the dental restoration for the user.

[0036] In some embodiments, the target restoration description text may include a description of the desired tooth color, such as pure white, off-white, or other colors. Because everyone's tooth color has subtle differences, or the same person's tooth color may vary at different stages, this is to better provide personalized services and ensure tooth color options. Specifically, in addition to direct descriptions like "pure white," the description text about tooth color can also be indirect, such as "the same tooth color as my current teeth." When the user's language expression ability is limited and the input description text about tooth color is relatively simple, text semantic understanding processing can also be performed on this type of indirect description language.

[0037] In this embodiment, the target dental model is a dental model obtained by scanning before generating the target dental restoration model. The target dental model includes the maxilla and mandible. The maxilla is located in the upper center of the face and is fixed, while the mandible is located in the lower part of the face and is movable. The occlusal relationship between the maxilla and mandible can be determined by scanning or retrieving historical data.

[0038] As an example, the description text of the target restoration corresponding to the design style of dental restorations can be "natural and like a real tooth", or "fits the morphological characteristics of adjacent teeth", or "round and full shape". Other design styles are not all listed here, and there are no specific limitations on the design style of dental restorations.

[0039] In dentistry, teeth are numbered using tooth position numbers. The most common system currently is the International Dental Federation (FDI) tooth position numbering system. The FDI tooth position number uses two digits. The first digit indicates the maxillofacial region and quadrant, and the second digit indicates the tooth's position within that quadrant. The oral cavity is divided into left and right sides by the midline, and further divided into upper and lower jaws, forming four quadrants. For the first digit: starting from the patient's right posterior position, the four quadrants are numbered clockwise from 1 to 4. Quadrant 1 represents the right maxillary region, quadrant 2 represents the left maxillary region, quadrant 3 represents the left mandibular region, and quadrant 4 represents the right mandibular region. For the second digit: starting from the midline (i.e., the incisors), the numbering proceeds backward from 1 to 8 or 1 to 5. As an example, for adults or minors who have already lost their primary teeth, each quadrant contains 8 teeth, numbered from 1 to 8. 1 represents the central incisor, 2 the lateral incisor, 3 the canine, 4 the first premolar, 5 the second premolar, 6 the first molar, 7 the second molar, and 8 the third molar (commonly known as the wisdom tooth). As an example, for children who haven't lost their primary teeth, each quadrant contains 5 teeth, numbered from 1 to 5. 1 represents the primary central incisor, 2 the primary lateral incisor, 3 the primary canine, 4 the first primary molar, and 5 the second primary molar. As an example, 11 represents the central incisor in the upper right jaw.

[0040] In some embodiments, the target dental model can be obtained through either the first or second method: The first method involves scanning the target user's mouth or plaster model using a target scanner, which may include an oral scanner or a fixed scanner. The second method involves obtaining the target dental model through a local database or a cloud database.

[0041] Among them, oral scanners are mainly used in dental treatments such as tooth restoration, implantation, or orthodontics. Oral scanners work based on optical principles. Doctors use handheld oral scanners to directly scan inside the patient's mouth to obtain three-dimensional data models of the teeth or gums.

[0042] Fixed scanners are permanently installed. As one example, they read pre-set barcode labels, obtain label information, and output the patient's target dental model information carried on the barcode label as 3D point cloud data. As another example, a fixed scanner, such as a desktop scanner, can scan a plaster model of the patient's target dental jaw to obtain a 3D data model of the teeth or gums.

[0043] In some embodiments, to simplify the treatment process, the target dental model is obtained directly from a local database or a cloud database. The target dental model is imported into the local database or cloud database in advance.

[0044] The computer device for generating dental prostheses is equipped with a trained model for generating dental prostheses. The trained model is used to process features from the target jaw model and the descriptive text of the target prosthesis, and then a model of the target dental prosthesis is generated.

[0045] The steps for training the prosthesis generation model are as follows: The first step involves acquiring multiple historical data sets. These historical data are then organized to create a sample set, which includes multiple training samples. This training set comprises four types of data: 3D dental prosthesis sample models, descriptive text sample data, intermediate dental restoration sample models, and target dental restoration sample models, all presented as quadruples. The text data includes morphological, functional, and aesthetic descriptions of the target dental restoration. If the historical data sets contain sensitive information such as patient privacy or trade secrets, they can be anonymized. Anonymization preserves the original data format and characteristics, only deleting, replacing, generalizing, or randomizing sensitive information to prevent identification of specific individuals or entities, while maintaining the data's other usability.

[0046] The second step involves the dental prosthesis generation device acquiring historical case data after desensitization processing. This data is then used to generate a sample set, which consists of four-tuples. These four-tuples include a 3D model of the jawbone, historically input text data, intermediately generated dental prosthesis models, and a target dental prosthesis model. This sample set is then divided into training and validation datasets according to a preset ratio.

[0047] The third step involves inputting the training dataset and validation dataset to the preset training model, iteratively training the training model based on the training dataset, and validating it based on the validation dataset to obtain a dental restoration generation model that meets the preset accuracy conditions.

[0048] S120. Input the target dental model, target tooth position number, and target restoration description text into the trained restoration generation model, perform feature extraction and fusion processing, and generate the target dental restoration model.

[0049] There is no specific limit to the number of target tooth positions; there can be one or more.

[0050] As an example, the description text for the target restoration could be "naturally similar to a real tooth", "fits the morphological features of adjacent teeth", or "rounded and full in shape", etc.

[0051] In some embodiments, the target dental prosthesis model, target tooth position number, and target prosthesis description text are input into the trained prosthesis generation model, and feature extraction and fusion processing is performed to generate a target dental prosthesis model and / or a target dental prosthesis model rendering. This can be achieved by generating a separate target dental prosthesis model for individual design result display, or by placing the generated target dental prosthesis model on the jawbone as a separate target dental prosthesis model rendering to achieve the overall effect, or by providing both a target dental prosthesis model and a target dental prosthesis model rendering for convenient user comparison. The rendering can be a two-dimensional or three-dimensional effect display.

[0052] In some embodiments, steps S120, such as Figure 2 This includes steps S1201-S1204: Step S1201: Input the target jaw model, target tooth position number, and target restoration description text into the restoration generation model; Step S1202: Based on the target jaw model and the target tooth position number, feature extraction is performed to obtain jaw space feature vector, tooth position number feature vector and tooth position spatial location feature vector; Feature extraction is performed based on the target jaw model and target tooth position number to obtain jaw space feature vector, tooth position number feature vector, and tooth position spatial location feature vector.

[0053] The dentitional spatial feature vector contains any one or a combination of morphological features, size information, spatial location information, and occlusal relationship information of the teeth associated with the target tooth position. The spatial location feature vector can be a feature vector of three-dimensional spatial coordinates. The teeth associated with the target tooth position include adjacent teeth, opposing teeth, and contralateral teeth of the same name.

[0054] Since the target dentition includes information such as tooth morphology, size, and occlusion, the dentition space feature vector obtained by feature extraction from the target dentition model can include information associated with the target tooth position number, such as the morphological information, size information, spatial position information, and occlusal relationship information of adjacent teeth; the morphological features, size information, spatial position information, and occlusal relationship information of opposing teeth; and the morphological information, size information, spatial position information, and occlusal relationship information of the contralateral corresponding tooth. At least one of the above information can be contained in the dentition space feature vector.

[0055] In some embodiments, since adjacent teeth are close to the target tooth position, the morphological, size, and spatial characteristics of the target tooth position can be determined through the morphological, size, and spatial position characteristics of the adjacent teeth. Since the opposing teeth bite the target tooth position longitudinally relative to it, the occlusal relationship characteristics of the target tooth position can be determined through the occlusal relationship characteristics of the opposing teeth. Since the contralateral tooth is located horizontally opposite the target tooth position, and the contralateral tooth generally has similar morphology and size to the target tooth position, and their spatial positions and occlusal relationships are also similar, the morphological, size, spatial position, and occlusal relationship characteristics of the target tooth position can be further determined through the contralateral tooth. This correction of the multi-dimensional spatial feature vector of the target tooth position using the contralateral tooth makes the dental restorations generated on the target tooth position in subsequent steps more accurate and better suited to the user's actual oral condition.

[0056] As an example, a point cloud of the target dental model is obtained by scanning with an oral scanner. Features are extracted from the point cloud using a point cloud network, and then the point cloud is converted into a sparse voxel mesh. The sparse voxel mesh contains local dental spatial feature vectors. The steps are as follows: The first step is to obtain a target dental model of the user's oral cavity by scanning with an oral scanner. The scanned data is point cloud data about the target dental model. The second step involves extracting features from the point cloud dataset corresponding to the target dentition model using the point cloud network in the geometric encoder. Feature learning is performed at the point level, with each point cloud interacting with its neighboring points to obtain the point cloud feature vector of the associated dentition of the target dentition (including any combination of adjacent teeth, opposing teeth, and contralateral teeth of the same name). This point cloud feature vector contains the contextual information of the local area surrounding the point. The third step is to first convert the point cloud into a three-dimensional voxel mesh using a geometric coding component, and then perform voxel space partitioning on the three-dimensional voxel mesh. Only non-empty three-dimensional voxels are stored to obtain a sparse voxel mesh. Each non-empty three-dimensional voxel in the voxel mesh is attached with a local jaw space feature vector of the associated jaw. The local jaw space feature vector is obtained based on the point cloud feature vector. The point cloud is voxelized using a geometric coding component, transforming the target dental model into a voxel representation. This process assigns the point cloud with feature vectors from the first step to the corresponding voxels. A voxel, short for "volume pixel," is a small cube in three-dimensional space that represents an attribute such as color, density, and material. The entire three-dimensional space can be filled with these uniformly spaced small cubes, forming a voxel mesh.

[0057] Voxel partitioning is a crucial step in computer graphics, medical imaging, and three-dimensional (3D) printing. By partitioning the space into voxels, complex 3D models or data can be efficiently represented, processed, analyzed, and interacted with. Whether the raw data comes from scans such as computed tomography (CT) or magnetic resonance imaging (MRI) to obtain the target dental model, or from 3D models drawn using CAD software, it is all processed in a unified voxel format for easier computer processing. Each voxel contains multiple points with point cloud feature vectors; these feature vectors are aggregated into a feature vector for that voxel by averaging or max pooling.

[0058] Only meaningful feature information is retained for the region occupied by the target jaw; the rest is empty and therefore does not need to be retained. This compresses the relatively dense 3D model representation into a sparse voxel representation. The geometric encoder encodes the scanned point cloud into a feature-rich sparse voxel representation, and this voxel implicitly contains feature information that can be used for geometric reconstruction. Utilizing sparsity, computation is only performed on non-empty voxels, significantly reducing memory and computational costs.

[0059] Specifically, starting randomly from any non-empty 3D voxel in the aforementioned 3D voxel grid, Gaussian noise is added to the 3D voxel to create random noise. Each non-empty 3D voxel is expanded into a voxel sequence unit, where each sequence unit contains one non-empty 3D voxel. Through feature concatenation, the initial text feature vector and the dental space feature vector are added as conditional feature vectors to the beginning of the voxel sequence unit. A time step is also added to the voxel sequence unit, which also includes a noise sequence.

[0060] A sequence of voxel units, incorporating conditional feature vectors, time steps, and noise data, is input into a Transformer-based geometric encoder. The Transformer's self-attention mechanism allows each voxel feature to interact with all other voxel features. Because of the linking at the local dentition space feature vector level, global contextual information is included, and the non-empty 3D voxels contain global dentition space feature vectors related to the dentition. To conform to the conditional feature vectors, denoising directions are predicted collaboratively across all voxel features to remove some noise. Thus, feature enhancement processing of the dentition space feature vectors using the Transformer network yields a feature-enhanced sparse voxel representation, a high-quality 3D voxel representation. The features carried by the sparse voxel representation constitute the intermediate dental restoration space feature vector, a high-dimensional feature vector encoding local surface orientation, curvature, and higher-level semantic information. Subsequent networks can decode the sparse voxel representation into a point cloud based on the intermediate dental restoration space feature vector. Finally, a fusion operation can be used to restore the sparse voxel representation in sequence unit form to a surface mesh, resulting in a digital model of the dental restoration.

[0061] Step S1203: Extract features from the target repair body description text to obtain the target text feature vector; The repair generation model includes a text encoder, which is used for text semantic understanding.

[0062] As an example, a text encoder for semantic understanding can include either Contrastive Language-Image Pre-training (CLIP) or Bidirectional Encoder Representations from Transformers (BERT)—either language understanding models. CLIP is used to process the deep semantics of images and text, while BERT has powerful bidirectional understanding capabilities, enabling it to grasp the precise meaning of words in specific contexts and the logic between sentences.

[0063] By processing the target restoration description text, such as "naturally shaped like a real tooth," using CLIP or BERT, an initial text feature vector based on the user's design expectations is obtained. The target text feature vector captures the semantics of the text, including design style and / or functional requirements.

[0064] Step S1204: Perform feature fusion processing on the jaw space feature vector, the tooth position number feature vector, the tooth position spatial location feature vector, and the target text feature vector to generate the target dental restoration model.

[0065] Feature fusion processing is performed on the dentition space feature vector, tooth position number feature vector, tooth position spatial location feature vector, and target text feature vector to generate the target dental restoration model.

[0066] like Figure 3 Step S1204 includes steps A1-A3: Step A1: Perform feature fusion processing on the tooth position number feature vector, tooth position spatial location feature vector, spatial feature vector of the previously generated dental restoration, jaw space feature vector, and target text feature vector to obtain the spatial feature vector of the target dental restoration. The feature vectors of tooth position number, tooth position spatial location, the spatial feature vector of the previously generated dental restoration, the jaw space feature vector, and the initial text feature vector are fused to obtain the spatial feature vector of the target dental restoration.

[0067] Step A2: Before feature fusion processing, determine whether the target restoration description text corresponding to the target text feature vector has been modified. If it has not been modified, set the spatial feature vector of the previously generated dental restoration to 0. When performing feature fusion processing based on the target restoration description text for the first time, the spatial feature vector of the previously generated dental restoration does not exist. Therefore, during the initial feature fusion processing, the spatial feature vector of the previously generated dental restoration is set to 0. Step A2 covers all steps of the feature fusion processing, including the initial feature fusion process and subsequent iterative fusion processes. Therefore, before feature fusion processing, it is determined whether the target restoration description text corresponding to the target text feature vector has been modified. If no modification is found, the spatial feature vector of the previously generated dental restoration is set to 0 to simulate the initial feature fusion process.

[0068] Step A3: Generate a model of the target dental restoration based on the spatial feature vector of the target dental restoration.

[0069] A model of the target dental restoration is generated based on the spatial feature vector of the target dental restoration.

[0070] The prosthesis generation model includes a geometric encoder, a geometric decoder, a text encoder, and a conditional diffusion module.

[0071] The geometric encoder encodes the point cloud of the target dental model into a sparse voxel representation with features. Only the regions occupied by the target dental restoration model retain meaningful feature information; unoccupied regions are empty and therefore do not need to be retained. The geometric encoder compresses the relatively dense 3D model representation into a relatively sparse voxel representation, and this voxel implicitly contains feature information that can be geometrically recovered.

[0072] The conditional diffusion module is used to fuse conditional feature vectors.

[0073] The geometry decoder is used to recover the geometric representation of the surface mesh from the compressed, sparse voxel representation.

[0074] like Figure 4 Step S1204 further includes steps B1-B2: Step B1: The spatial feature vector of the target dental restoration is obtained by performing feature fusion processing on the dental space feature vector, the tooth position number feature vector, the tooth position spatial location feature vector, and the target text feature vector through the conditional diffusion module and the geometric encoder. Step B2: Generate a model of the target dental restoration based on the spatial feature vector of the target dental restoration using a geometric decoder.

[0075] The sparse voxel representation with features output from the aforementioned geometric encoder is used as the intermediate dental restoration spatial feature vector and input into the geometric decoder. The geometric decoder decodes the intermediate dental restoration spatial feature vector to obtain the target dental restoration model. The geometric decoder decodes the continuous dense surface of the object implied by the sparse voxels and outputs a high-resolution explicit triangular mesh.

[0076] If the target dental model input to the geometry encoder is in point cloud form, then the target dental restoration model output by the geometry decoder is also in point cloud form.

[0077] like Figure 5 Step B2 includes steps C1-C2: Step C1: Input the spatial feature vector of the target dental restoration into the geometric decoder for decoding processing to obtain the intermediate dental restoration model; As an example, the geometric decoder can be a 3D convolutional neural network model. It inputs the sparse voxel coordinates and features corresponding to the spatial feature vector of the intermediate dental restoration into the 3D convolutional neural network model, improving spatial resolution through sparse transpose convolution or sparse interpolation operations. The geometric decoder uses the features of adjacent voxels in the sparse voxels to fill in the features of newly added voxels to complete the spatial geometric information, ultimately outputting a target dental restoration model, which is a dense voxel mesh.

[0078] The method described above, which outputs the target dental restoration model in the form of a dense voxel mesh, can recover the target dental restoration model, but it consumes a huge amount of memory.

[0079] Furthermore, for the target dental restoration model that outputs a dense voxel mesh using the aforementioned memory-intensive method, a different approach can be taken: instead of outputting a dense voxel mesh, the dense voxel mesh can be used as an implicit function to save memory. Multiple non-empty voxels near a predefined query voxel point are found in the sparse voxel feature field. Based on an attention mechanism, a feature related to the query voxel point is extracted from these nearby non-empty voxels. The occupancy probability of the query voxel point is predicted based on the extracted feature. Finally, a continuous function is output. This continuous function allows querying the geometric state of any voxel in voxel space, generating one or more triangular facets within the voxel to approximate the surface of the target dental restoration.

[0080] In some embodiments, the generated target dental restoration model is stored in a local database and may not be displayed. If an external device is received requesting the target dental restoration model, the model is sent to that device. Alternatively, the model can be displayed to the user, indicating that the target dental restoration has been designed. At this point, a target dental restoration with the user's desired style has been designed, and the user can fabricate the dental prosthesis based on this model.

[0081] Step C2: When a user's confirmation instruction for the intermediate dental prosthesis model is received, or when no user's modification instruction for the intermediate dental prosthesis model is received within a preset time period, the intermediate dental prosthesis model is used as the target dental prosthesis model.

[0082] In some embodiments, the intermediate dental restoration model generated above is based on the text output once. The user may have new expectations for the intermediate dental restoration model. Therefore, the dental restoration model generated once can be used as the intermediate dental restoration model. After multiple interactions with the user in the form of text commands, the final dental restoration model is generated. In this way, the input text is modified multiple times, and the target model is gradually modified so that the target dental restoration model is more in line with the user's needs. When the user's confirmation command for the intermediate dental restoration model is received, or when no modification command for the intermediate dental restoration model is received from the user within a preset time period, the intermediate dental restoration model is used as the target dental restoration model.

[0083] like Figure 6 Step B2 also includes steps D1-D6: Step D1: Input the spatial feature vector of the target dental restoration into the geometric decoder for decoding processing to obtain the intermediate dental restoration model; The spatial feature vector of the target dental restoration is input into the geometric decoder for decoding to obtain the intermediate dental restoration model.

[0084] Step D2: When a user's modification instruction for the first dental prosthesis model of the intermediate dental prosthesis model is received within a preset time period, the text feature extraction process is performed on the modification description text in the modification instruction of the first dental prosthesis model through a text encoder to obtain the modification text feature vector. If the intermediate dental restoration model generated in one go receives a confirmation instruction from the user by clicking the confirmation item in the dialog box after being displayed, then the intermediate dental restoration model will be used as the target dental restoration model.

[0085] A preset time period is designated for user confirmation or modification. Within this time period, the user can confirm the dental restoration model. If the user confirms the generated intermediate dental restoration model within the preset time period, this intermediate model will be used as the target dental restoration model. If the user does not confirm within the preset time period, it may be because the user has not yet decided whether to confirm or not, or the user has decided not to confirm but to modify. In this case, a new dental restoration model will be generated based on the user's revised description of the dental restoration design style. If no modification instruction for the intermediate dental restoration model is received from the user within the preset time period, this intermediate model will be used as the target dental restoration model.

[0086] As an example, user-inputted modification text might include phrases like "deeper the occlusal grooves," "make the axial profile more rounded," or "reduce the volume by 5%."

[0087] Step D3: The dental restoration is adjusted by the conditional diffusion module and the geometric encoder based on the modified text feature vector, the jaw space feature vector, and the intermediate dental restoration space feature vector corresponding to the intermediate dental restoration model, so as to obtain the dental restoration adjustment space feature vector. The conditional diffusion module, based on the modified text feature vector, the jaw space feature vector obtained in the previous step, and the intermediate dental restoration space feature vector corresponding to the intermediate dental restoration model, performs dental restoration adjustment processing to obtain the dental restoration adjustment space feature vector. Furthermore, the dental restoration adjustment space feature vector can be further determined based on the tooth position feature vector and the tooth spatial location feature vector.

[0088] In some embodiments, a user can input modification commands to the first dental prosthesis model of the intermediate dental prosthesis model multiple times. Each time a spatial feature vector for adjusting the dental prosthesis is generated based on the user's modification commands, the intermediate dental prosthesis spatial feature vector is used. Only when the user initially inputs descriptive text expressing the desired style of the dental prosthesis is the intermediate dental prosthesis spatial feature vector not used, because the initial descriptive text is not the modification text, and the intermediate dental prosthesis feature vector has not yet been generated. Therefore, the aforementioned step of generating the intermediate dental prosthesis model for the first time is improved by adding a separate variable specifically for the intermediate dental prosthesis feature vector. This variable stores the value of the intermediate dental prosthesis feature vector, thus unifying the steps of generating the first dental prosthesis model with subsequent model modification steps. Only the value of this variable needs to be set to 0 in the first step of generating the dental prosthesis model. In subsequent model modification iterations, this variable records the intermediate dental prosthesis feature vector. By setting such a variable, only the value in the variable needs to be extracted for modification, instead of regenerating a dental prosthesis model that conforms to the modification commands from scratch. Therefore, modifying the dental prosthesis model generated earlier, which may not fully meet the user's expectations, can reduce the amount of computation, speed up model generation, and further improve the design efficiency of dental prosthesis models.

[0089] Step D4: Generate a dental restoration adjustment model based on the geometric decoder and the spatial feature vector of the dental restoration adjustment. This section describes the steps for generating an adjustment model of a dental prosthesis based on a geometric decoder and the spatial feature vectors of the dental prosthesis adjustment space. Boolean operations, such as intersection or OR operations, are performed on the 3D voxels using the spatial feature vectors to adjust the intermediate dental prosthesis model, resulting in the adjusted model. This model is generated based on a secondary text description input by the user. Adjustments are made immediately based on deep understanding text commands, allowing the user to instantly modify or confirm the adjusted model. This shortens the interaction cycle between the deep understanding text-based dental prosthesis model and the user, accelerating the user confirmation process and thus improving the efficiency of dental prosthesis design. The improved efficiency is due to three main factors: first, the automatic model generation replaces the manual modeling process in CAD software, resulting in faster automation; second, the timely delivery of design results to the user shortens the interaction cycle; and third, the use of a deep language understanding model based on user big data analysis, which may offer superior understanding compared to manual analysis. The dental prosthesis model generated by this text encoder has a higher accuracy in reproducing the user's desired model, reducing the number of revisions. In summary, this improves the efficiency of dental prosthesis design.

[0090] Step D5: When a user confirms the adjustment model of the dental prosthesis, or when no user modifies the adjustment model of the dental prosthesis is received within a preset time period, the adjustment model of the dental prosthesis is used as the target dental prosthesis model. In some embodiments, if the user is satisfied with the first modification to the dental prosthesis model, the user can click the confirmation button in the dialog box to generate a confirmation instruction for adjusting the dental prosthesis model.

[0091] Give the user a preset confirmation time. If the user does not confirm within the preset time period, the dental restoration adjustment model can still be used as the target dental restoration model. Alternatively, if no modification text is received from the user within the preset time period, the dental restoration adjustment model can still be used as the target dental restoration model.

[0092] Step D6: When a user's modification instruction for the second dental prosthesis model is received within a preset time period, the dental prosthesis adjustment model is updated to an intermediate dental prosthesis model, the modification instruction for the second dental prosthesis model is updated to the modification instruction for the first dental prosthesis model, and the process of extracting text features from the modification description text in the modification instruction for the first dental prosthesis model using a text encoder is returned to execute the step of obtaining the modification text feature vector.

[0093] If a user's modification instruction for the first dental prosthesis model is received within a preset 5-minute timeframe, the text feature extraction process is performed on the modification description text in the modification instruction using a text encoder. This text feature extraction process is similar to the first text extraction method described above, resulting in a modified text feature vector.

[0094] If a modification instruction is received from the user within a preset time period, and the user is not satisfied with the model after the first modification, the user inputs modification text for a second dental restoration model to adjust the model. Upon receiving the modification instruction, the dental restoration generation device adjusts the model again based on the instruction. The modification instruction for the second dental restoration model is updated to the modification instruction for the first dental restoration model. The process then returns to the step of extracting text features from the dental restoration style modification description text in the first dental restoration model modification instruction using a text encoder to obtain the modification text feature vector.

[0095] In some embodiments, after step S120, the method further includes: sending the target dental prosthesis model to a preset user terminal, or sending the target dental prosthesis model to a 3D printing device for the 3D printing device to print based on the selected material.

[0096] In some embodiments, after the target dental restoration model is generated, the target dental restoration can be sent to the user terminal for display if needed.

[0097] In some embodiments, after generating the target dental restoration model, the target dental restoration model can be sent to a 3D printing device for printing based on a selected material. The 3D printing device can be a 3D printer or a connected computing device, and the selected material can be resin or other printing materials.

[0098] This application provides a method, apparatus, computer device, and storage medium for generating dental prostheses. The method includes: acquiring a target jaw model, a target tooth position number, and a target prosthesis description text, where the target tooth position number corresponds to the tooth position of the target tooth in the dental prosthesis to be designed; inputting the target jaw model, target tooth position number, and target prosthesis description text into a trained prosthesis generation model for feature extraction and fusion processing to generate a target dental prosthesis model. This application directly generates a target dental prosthesis model based on the user's description text and the target jaw model, replacing the step of relying on manual experience for understanding and modeling. It directly automates semantic understanding and output, deeply understands the user's design expectations, simplifies the modeling process, and solves the technical problem of low efficiency in dental prosthesis design in the prior art.

[0099] Figure 7This is a schematic block diagram of a dental prosthesis generation device provided in an embodiment of this application. Figure 7 As shown, corresponding to the above-described method for fabricating dental prostheses, this application also provides a dental prosthesis fabrication apparatus 600. This dental prosthesis fabrication apparatus 600 includes a unit for performing the above-described method for fabricating dental prostheses, and can be configured in a terminal such as a desktop computer, tablet computer, or laptop computer. Specifically, please refer to... Figure 7 The dental prosthesis generation device 600 includes an acquisition unit 601 and a model generation unit 602, wherein: The acquisition unit 601 is used to acquire the target jaw model, the target tooth position number and the target restoration description text, wherein the target tooth position number is the tooth position number corresponding to the target tooth position of the dental restoration to be designed. The model generation unit 602 is used to perform feature extraction and fusion processing on the target dental prosthesis model, the target tooth position number, and the target prosthesis description text input into the trained prosthesis generation model to generate the target dental prosthesis model.

[0100] In some embodiments, the dental prosthesis generation method is applied to a dental prosthesis generation computer device, which deploys a trained dental prosthesis generation model, including a geometric encoder, a geometric decoder, a text encoder, and a conditional diffusion module.

[0101] In some embodiments, the model generation unit 602 performs feature extraction and fusion processing on a prosthesis generation model trained by inputting the target jaw model, target tooth position number, and target prosthesis description text to generate a target dental prosthesis model, specifically for: Input the target jaw model, target tooth position number, and target restoration description text into the restoration generation model; Based on the target jaw model and target tooth position number, feature extraction is performed to obtain jaw space feature vector, tooth position number feature vector and tooth position spatial location feature vector; Feature extraction is performed on the description text of the target repair body to obtain the target text feature vector; Feature fusion processing is performed on the dentition space feature vector, tooth position number feature vector, tooth position spatial location feature vector, and target text feature vector to generate the target dental restoration model.

[0102] In some embodiments, the model generation unit 602 performs feature fusion processing on the jaw space feature vector, tooth position number feature vector, tooth position spatial location feature vector, and target text feature vector to generate a target dental restoration model, specifically for: The feature fusion processing is performed on the tooth position number feature vector, the tooth position spatial location feature vector, the spatial feature vector of the previously generated dental restoration, the jaw space feature vector, and the target text feature vector to obtain the spatial feature vector of the target dental restoration. Before feature fusion processing, it is determined whether the target restoration description text corresponding to the target text feature vector has been modified. If it has not been modified, the spatial feature vector of the previously generated dental restoration is set to 0. A model of the target dental restoration is generated based on the spatial feature vector of the target dental restoration.

[0103] In some embodiments, the model generation unit 602 performs feature extraction and fusion processing on the prosthesis generation model trained by inputting the target jaw model, target tooth position number, and target prosthesis description text to generate a target dental prosthesis model. The prosthesis generation model includes a geometric encoder, a geometric decoder, a text encoder, and a conditional diffusion module.

[0104] In some embodiments, the model generation unit 602 performs feature fusion processing on the jaw space feature vector and the initial text feature vector to generate a target dental restoration model, specifically for: The spatial feature vector of the target dental restoration is obtained by using the conditional diffusion module and the geometric encoder to perform feature fusion processing on the tooth position number feature vector and the tooth position spatial location feature vector, the spatial feature vector of the previously generated dental restoration, the jaw space feature vector, and the initial text feature vector. A model of the target dental restoration is generated based on the spatial feature vector of the target dental restoration using a geometric decoder.

[0105] In some embodiments, the model generation unit 602 generates a model of the target dental restoration based on the spatial feature vector of the target dental restoration using a geometry decoder, specifically for: The spatial feature vector of the target dental restoration is input into the geometric decoder for decoding to obtain the intermediate dental restoration model. When a user confirms the intermediate dental prosthesis model, or when no user modifies the intermediate dental prosthesis model is received within a preset time period, the intermediate dental prosthesis model will be used as the target dental prosthesis model.

[0106] In some embodiments, the model generation unit 602 generates a model of the target dental restoration based on the spatial feature vector of the target dental restoration using a geometry decoder, specifically for: The spatial feature vector of the target dental restoration is input into the geometric decoder for decoding to obtain the intermediate dental restoration model. When a user's modification instruction for the first dental prosthesis model is received within a preset time period, the text feature extraction process is performed on the modification description text in the modification instruction for the first dental prosthesis model through a text encoder to obtain the modification text feature vector. The dental restoration is adjusted by the conditional diffusion module and the geometric encoder based on the modified text feature vector, the jaw space feature vector and the intermediate dental restoration space feature vector corresponding to the intermediate dental restoration model, and the dental restoration adjustment space feature vector is obtained. A dental restoration adjustment model is generated based on the geometric decoder and the feature vector of the dental restoration adjustment space. When a user confirms the adjustment model of the dental prosthesis, or when no user modifies the adjustment model of the dental prosthesis is received within a preset time period, the dental prosthesis adjustment model is used as the target dental prosthesis model. When a user's modification instruction for the second dental prosthesis model is received within a preset time period, the dental prosthesis adjustment model is updated to an intermediate dental prosthesis model, the modification instruction for the second dental prosthesis model is updated to a modification instruction for the first dental prosthesis model, and the process returns to the step of extracting text features from the modification description text in the modification instruction for the first dental prosthesis model using a text encoder to obtain the modification text feature vector.

[0107] In some embodiments, the generation model unit 602 performs feature extraction and fusion processing on the restoration generation model after inputting the target dental model, target tooth position number, and target restoration description text into the training, to generate a target dental restoration model. In this model, the dental spatial feature vector includes at least one of the following features: morphological features, size features, spatial position features, and occlusal relationship features of the associated teeth of the target tooth position. The associated teeth include the adjacent teeth, opposing teeth, and contralateral teeth of the same name of the target tooth position.

[0108] In some embodiments, when the acquisition unit 601 performs the acquisition of the target jaw model, the target tooth position number, and the target restoration description text, wherein the target tooth position number is the tooth position number corresponding to the target tooth position of the dental restoration to be designed, the target jaw model is obtained in the following manner: Obtained by scanning the oral cavity or plaster model of the target user using a target scanner, which may include an oral scanner or a fixed scanner; Alternatively, the target dental model can be obtained through a local database or a cloud database.

[0109] In some embodiments, after the model generation unit 602 generates a target dental restoration model based on the jaw space feature vector and the initial text feature vector, the dental restoration generation device 600 further includes: The sending unit 603 is used to send the target dental restoration model to a preset user terminal, or to send the target dental restoration model to a 3D printing device for the 3D printing device to print based on the selected material.

[0110] In some embodiments, the model generation unit 602 performs feature extraction and fusion processing on a prosthesis generation model trained by inputting the target jaw model, target tooth position number, and target prosthesis description text to generate a target dental prosthesis model, specifically for: The target dental prosthesis model, target tooth position number, and target prosthesis description text are input into the trained prosthesis generation model. Feature extraction and fusion processing are then performed to generate the target dental prosthesis model and / or the target dental prosthesis model rendering.

[0111] In summary, the dental prosthesis generation device 600 in this embodiment acquires a target jaw model, a target tooth position number, and a target prosthesis description text. The target tooth position number corresponds to the tooth position of the target tooth in the dental prosthesis to be designed. The target jaw model, target tooth position number, and target prosthesis description text are input into a trained prosthesis generation model for feature extraction and fusion processing to generate a target dental prosthesis model. This embodiment directly generates a target dental prosthesis model based on the user's description text and the target jaw model, replacing the step of relying on manual experience for understanding and modeling. It directly automates semantic understanding and output, deeply understands the user's design expectations, simplifies the modeling process, and solves the technical problem of low efficiency in dental prosthesis design in the prior art.

[0112] It should be noted that those skilled in the art can clearly understand that the specific implementation process of the above-mentioned dental prosthesis generation device and each unit can be referred to the corresponding description in the foregoing method embodiments. For the sake of convenience and brevity, it will not be repeated here.

[0113] The aforementioned dental prosthesis generation device can be implemented as a computer program, which can, for example... Figure 8 It runs on the computer device shown.

[0114] Please see Figure 8 , Figure 8 This is a schematic block diagram of a computer device 700 provided in an embodiment of this application. The computer device 700 can be a terminal or a server. The terminal can be an electronic device with communication functions, such as a smartphone, tablet, laptop, desktop computer, personal digital assistant, or wearable device. The server can be a standalone server or a server cluster composed of multiple servers.

[0115] See Figure 8The computer device 700 includes a processor 702, a memory, and a network interface 705 connected via a system bus 701. The memory may include a non-volatile storage medium 703 and internal memory 704.

[0116] The non-volatile storage medium 703 may store an operating system 7031 and a computer program 7032. The computer program 7032 includes program instructions that, when executed, cause the processor 702 to perform a method for generating a dental prosthesis.

[0117] The processor 702 provides computing and control capabilities to support the operation of the entire computer device 700.

[0118] The internal memory 704 provides an environment for the operation of the computer program 7032 in the non-volatile storage medium 703. When the computer program 7032 is executed by the processor 702, the processor 702 can execute a method for generating a dental prosthesis.

[0119] This network interface 705 is used for network communication with other devices. Those skilled in the art will understand that... Figure 7 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device 700 to which the present application is applied. The specific computer device 700 may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0120] The processor 702 is used to run the computer program 7032 stored in the memory to perform the following steps: Obtain the target jaw model, target tooth position number, and target restoration description text. The target tooth position number is the tooth position number corresponding to the target tooth position of the dental restoration to be designed. The target dental prosthesis model is generated by inputting the target jaw model, target tooth position number, and target prosthesis description text into the trained prosthesis generation model and performing feature extraction and fusion processing.

[0121] It should be understood that in the embodiments of this application, the processor 702 may be a central processing unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0122] It will be understood by those skilled in the art that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program includes program instructions and can be stored in a storage medium, which is a computer-readable storage medium. The program instructions are executed by at least one processor in the computer system to implement the process steps of the embodiments of the above methods.

[0123] Therefore, this application also provides a storage medium. This storage medium can be a computer-readable storage medium. The storage medium stores a computer program, wherein the computer program includes program instructions. When executed by a processor, the program instructions cause the processor to perform the following steps: Obtain the target jaw model, target tooth position number, and target restoration description text. The target tooth position number is the tooth position number corresponding to the target tooth position of the dental restoration to be designed. The target dental prosthesis model is generated by inputting the target jaw model, target tooth position number, and target prosthesis description text into the trained prosthesis generation model and performing feature extraction and fusion processing.

[0124] The storage medium can be any computer-readable storage medium that can store program code, such as a USB flash drive, external hard drive, read-only memory (ROM), magnetic disk, or optical disk.

[0125] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this application.

[0126] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative. For example, the division of each unit is merely a logical functional division, and there may be other division methods in actual implementation. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed.

[0127] The steps in the methods of this application embodiment can be adjusted, merged, or deleted according to actual needs. The units in the apparatus of this application embodiment can be merged, divided, or deleted according to actual needs. Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0128] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, a terminal, or a network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application.

[0129] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for manufacturing dental restorations, characterized in that, include: Obtain the target dental model, target tooth position number, and target restoration description text, wherein the target tooth position number is the tooth position number corresponding to the target tooth position of the dental restoration to be designed; The target dental prosthesis model is generated by performing feature extraction and fusion processing on the target jaw model, the target tooth position number, and the target prosthesis description text as inputs to the trained prosthesis generation model.

2. The method according to claim 1, characterized in that, The step of extracting and fusing features from the target dental prosthesis model, the target tooth position number, and the target prosthesis description text, and then inputting them into the trained prosthesis generation model, to generate the target dental prosthesis model, includes: Input the target dental model, the target tooth position number, and the target restoration description text into the trained restoration generation model; Based on the target dental model and the target tooth position number, feature extraction is performed to obtain the dental spatial feature vector, the tooth position number feature vector, and the tooth position spatial location feature vector. Feature extraction is performed on the description text of the target repair body to obtain the target text feature vector; The feature fusion processing is performed on the jaw space feature vector, the tooth position number feature vector, the tooth position spatial location feature vector, and the target text feature vector to generate the target dental restoration model.

3. The method according to claim 2, characterized in that, The step of performing feature fusion processing on the jaw space feature vector, the tooth position number feature vector, the tooth position spatial location feature vector, and the target text feature vector to generate the target dental restoration model includes: The feature fusion processing is performed on the tooth position number feature vector, the tooth position spatial location feature vector, the spatial feature vector of the previously generated dental restoration, the jaw space feature vector, and the target text feature vector to obtain the spatial feature vector of the target dental restoration. Before the feature fusion process, it is determined whether the target restoration description text corresponding to the target text feature vector has been modified. If there is no modification, the spatial feature vector of the previously generated dental restoration is set to 0. A model of the target dental restoration is generated based on the spatial feature vector of the target dental restoration.

4. The method according to claim 2, characterized in that, The repair body generation model includes a geometric encoder, a geometric decoder, a text encoder, and a conditional diffusion module.

5. The method according to claim 4, characterized in that, The step of performing feature fusion processing on the jaw space feature vector, the tooth position number feature vector, the tooth position spatial location feature vector, and the target text feature vector to generate the target dental restoration model includes: The spatial feature vector of the target dental restoration is obtained by performing feature fusion processing on the jaw space feature vector, the tooth position number feature vector, the tooth position spatial location feature vector, and the target text feature vector through the conditional diffusion module and the geometric encoder. The geometric decoder generates a model of the target dental restoration based on the spatial feature vector of the target dental restoration.

6. The method according to claim 5, characterized in that, The step of generating a target dental restoration model based on the spatial feature vector of the target dental restoration using the geometric decoder includes: The spatial feature vector of the target dental restoration is input into the geometric decoder for decoding to obtain an intermediate dental restoration model. When a user confirms the intermediate dental prosthesis model, or when no user modifies the intermediate dental prosthesis model is received within a preset time period, the intermediate dental prosthesis model is used as the target dental prosthesis model.

7. The method according to claim 5, characterized in that, The step of generating a target dental restoration model based on the spatial feature vector of the target dental restoration using the geometric decoder includes: The spatial feature vector of the target dental restoration is input into the geometric decoder for decoding to obtain an intermediate dental restoration model. When a user receives a modification instruction for the first dental prosthesis model of the intermediate dental prosthesis model within a preset time period, the text encoder performs text feature extraction processing on the modification description text in the modification instruction for the first dental prosthesis model to obtain the modification text feature vector as the target text feature vector. The dental restoration is adjusted by the conditional diffusion module and the geometric encoder based on the modified text feature vector, the dentition space feature vector and the intermediate dental restoration space feature vector corresponding to the intermediate dental restoration model, to obtain the dental restoration adjustment space feature vector. A dental prosthesis adjustment model is generated based on the geometric decoder and the dental prosthesis adjustment space feature vector. When a user confirms the adjustment model of the dental prosthesis, or when no user modifies the adjustment model of the dental prosthesis is received within a preset time period, the adjustment model of the dental prosthesis is used as the target dental prosthesis model. When a user's second dental prosthesis model modification instruction is received within a preset time period, the dental prosthesis adjustment model is updated to an intermediate dental prosthesis model, the second dental prosthesis model modification instruction is updated to a first dental prosthesis model modification instruction, and the process of extracting text features from the modification description text in the first dental prosthesis model modification instruction using the text encoder to obtain the modification text feature vector is returned.

8. The method according to claim 1, characterized in that, The dentition space feature vector includes at least one of the following features: morphological features, size features, spatial position features, and occlusal relationship features of the associated teeth of the target tooth position. The associated teeth include the adjacent teeth, opposing teeth, and contralateral teeth of the same name of the target tooth position.

9. The method according to claim 1, characterized in that, The target dental model was obtained through the following methods: Obtained by scanning the oral cavity or plaster model of the target user using a target scanner, wherein the target scanner includes an oral scanner or a fixed scanner; Alternatively, the target dental model can be obtained through a local database or a cloud database.

10. The method according to claim 1, characterized in that, After generating the target dental restoration model based on the jaw space feature vector and the initial text feature vector, the method further includes: Send the target dental restoration model to a preset user terminal; Alternatively, the target dental prosthesis model can be sent to a 3D printing device for printing based on the selected material.

11. The method according to claim 1, characterized in that, The step of extracting and fusing features from the target dental prosthesis model, the target tooth position number, and the target prosthesis description text, and then inputting them into the trained prosthesis generation model, to generate the target dental prosthesis model, includes: The target dental prosthesis model, the target tooth position number, and the target prosthesis description text are input into the trained prosthesis generation model. Feature extraction and fusion processing are then performed to generate the target dental prosthesis model and / or the target dental prosthesis model rendering.

12. A dental prosthesis generation device, characterized in that, The dental prosthesis generating device includes: The acquisition unit is used to acquire the target dental model, the target tooth position number, and the target restoration description text, wherein the target tooth position number is the tooth position number corresponding to the target tooth position of the dental restoration to be designed. The model generation unit is used to perform feature extraction and fusion processing on the target dental prosthesis model, the target tooth position number, and the target prosthesis description text input into the trained prosthesis generation model to generate the target dental prosthesis model.

13. A computer device for generating dental prostheses, characterized in that, The method includes a memory, a processor, and a dental prosthesis generation program stored in the memory and executable on the processor, wherein the processor executes the dental prosthesis generation program to implement the steps of the dental prosthesis generation method according to any one of claims 1 to 11.

14. A storage medium, characterized in that, The storage medium stores a program for implementing a method for generating a dental prosthesis, the program for implementing a method for generating a dental prosthesis being executed by a processor to implement the steps of the method for generating a dental prosthesis as described in any one of claims 1 to 11.

Images