A digital and intelligent denture manufacturing method

Abstract

The application discloses a digital and intelligent denture manufacturing method, relates to the technical field of artificial intelligence, and solves the problems of inconsistent quality standards of artificial production, uncontrollable production timeliness of dentures, low production precision, low production efficiency and difficult product improvement; the denture manufacturing method comprises the following steps: taking a mold; digitally importing; generating an AI model; AI tooth arrangement; digitally producing a trial denture; denture trial; guide plate design; digital guide plate replication and tooth arrangement; and finished product; through an improved PointNet++ algorithm and an improved denture production process, digital technology is utilized to realize standardized and streamlined denture production; through the establishment of a large database for storing and analyzing patient denture shape data, the denture production efficiency and denture delivery efficiency are improved, the denture processing time is reduced, and the comfort of the denture is improved.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of artificial intelligence, and in particular to a digital and intelligent denture manufacturing method. BACKGROUND

[0002] In recent years, with the continuous development of computer science and digital technology, the digital and intelligent denture manufacturing technology has gradually matured. The digital and intelligent denture manufacturing is a manufacturing process based on digital technology, which quickly and efficiently manufactures personalized dental crowns, removable partial dentures, complete dentures and other oral prostheses that meet the oral morphology and functional requirements of patients through steps such as digital scanning, simulation design and numerical control processing.

[0003] Patent No. CN2019102912014 discloses a denture and a production process thereof, and the technical scheme points of the application are as follows: the denture comprises porcelain teeth and a fixing part; the fixing part comprises two fixing pieces covering the top of two adjacent teeth, a connecting piece fixedly connected to the side of the two fixing pieces close to the porcelain teeth, and a plug-in piece fixedly connected to the bottom of the end of the connecting piece away from the porcelain teeth and vertically arranged; the porcelain teeth comprise a metal inner crown and a ceramic outer crown covering the outer side of the metal inner crown; the production process comprises the following steps: a. model trimming; b. wax type making; c. embedding; d. casting; e. polishing; f. porcelain wax type making; g. casting; h. porcelain application and glazing; and i. polishing. When the denture is installed, the two plug-in holes on the porcelain teeth are aligned with the two plug-in blocks, so that the plug-in blocks are inserted into the plug-in holes, thereby fixing the porcelain teeth and the two adjacent teeth, and when the denture is disassembled, the porcelain teeth are pushed upward so that the plug-in blocks and the plug-in holes are separated from each other, and the porcelain teeth are convenient to disassemble and clean.

[0004] Patent No. CN202310151595X discloses a full-ceramic denture, which comprises a denture and a tooth root. The denture and the tooth root are connected by mutual adhesion and nesting through a fixing agent. The bottom surface of the denture is provided with an adhesion groove. The adhesion groove is provided with a reinforced adhesion groove in the middle and an adhesion inner groove in the middle of the reinforced adhesion groove. Through the structural combination design of the flushing mechanism and the drying mechanism on the other side of the processing equipment, rapid high-pressure flushing of multiple sets of polished full-ceramic dentures and double rapid drying of wind energy and heat energy after flushing are realized, thereby realizing efficient production operation. Through the above reasonable structure design, the operation is simple and convenient, the production is smooth and efficient, time and labor are saved, and good economic benefits are expected to be achieved.

[0005] Although the above patents process and produce dentures through traditional manufacturing processes, due to the inconsistent quality standards of manual production at present, the production timeliness of the dentures cannot be controlled, the production precision is not high, the production efficiency is low, and product improvement is difficult. SUMMARY

[0006] The application aims to provide a digital denture manufacturing method, which can realize standardized and streamlined denture production by using digital technology and improving the PointNet++ algorithm and the denture production process, and can also improve the denture production efficiency and delivery efficiency, reduce the processing time of the denture, and improve the comfort of the denture by establishing a large database to store and analyze the denture morphology data of patients.

[0007] The application uses the following technical solutions:

[0008] A digital denture manufacturing method comprises the following steps:

[0009] S1: Obtain and store the anatomical structure data of the patient's oral cavity;

[0010] The anatomical structure data includes two-dimensional data and point cloud data; the point cloud data includes alveolar ridge point cloud data, hard palate point cloud data, edentulous jaw point cloud data, mastication feature point cloud data, and alveolar arch point cloud data; the two-dimensional data includes condyle path inclination, maximum occlusal clearance, normal occlusal clearance, and intermaxillary distance;

[0011] S2: Data cleaning is performed on the anatomical structure data, and the AI design software is imported;

[0012] S3: The AI design software generates an intraoral alveolar bone model according to the cleaned anatomical structure data, and performs finite element analysis on the intraoral alveolar bone model;

[0013] S4: The computer-aided denture arrangement method or the improved PointNet++ algorithm is used in the intraoral alveolar bone model to arrange the denture according to the anatomical structure data, and the arranged digital denture is extracted, and then a denture model is obtained, and a bracket and / or a steel support are added to the denture model according to the patient's needs;

[0014] S5: The denture model is used for digital production of trial dentures by using 3D printing cutting equipment to provide the patient with trial dentures; the denture model includes a complete denture model and a fixed denture model;

[0015] S6: The three-dimensional data of the denture model is corrected according to the feedback of the patient's trial denture to obtain the design data of the final denture model;

[0016] S7: The AI design software generates final denture guide plate data according to the design data of the final denture model, and uses the 3D printing cutting equipment to manufacture the final denture guide plate according to the final denture guide plate data;

[0017] S8: The production personnel complete the denture arrangement, shape fixation and detail adjustment according to the final denture guide plate to determine the final denture model; the detail adjustment includes polishing, polishing and cleaning;

[0018] S9: using digital technology and traditional technology, the final denture model is used to manufacture the finished product.

[0019] Preferably, in step S1, the doctor uses an intraoral scanning device to accurately scan the inside of the patient's mouth, or uses a traditional mold taking process to obtain a patient's intraoral model, and then uses a scanner to scan the model to obtain the patient's anatomical structure data; the anatomical structure data of the patient's mouth is stored and analyzed through a large database.

[0020] Preferably, step S2 includes the following steps:

[0021] S2-1: The point cloud data in the anatomical structure data is adaptively voxelized to form a plurality of three-dimensional voxel grids; the two-dimensional data in the anatomical structure data is denoised using a data cleaning algorithm to obtain denoised two-dimensional data;

[0022] S2-2: Calculate the center of mass and the center of the three-dimensional voxel grid to obtain the center of mass voxel block and the center voxel block, respectively;

[0023] S2-3: Using a Gaussian function, the spatial domain kernel of the center of mass voxel block is calculated according to the Euclidean distance of the point cloud position, and the spatial domain kernel is subjected to Fourier transform to obtain a spatial domain matrix;

[0024] S2-4: Using a multi-logarithmic function, the value domain kernel of the center voxel block is calculated according to the gray similarity, and the value domain kernel is subjected to fast Fourier transform to obtain a value domain matrix;

[0025] S2-5: Multiply the spatial domain matrix and the value domain matrix to obtain denoised point cloud data;

[0026] S2-6: The denoised point cloud data and the denoised two-dimensional data are transmitted to the AI design software using a parallel transmission algorithm.

[0027] Preferably, in step S3, the AI design software first correlates the anatomical structure data according to the oral structure using a data correlation algorithm to generate an oral structure data table; then generates a virtual oral model according to the oral structure data table using a three-dimensional reconstruction algorithm; subsequently, the virtual oral model is edge enhanced using a Gaussian Laplace algorithm, and the outline traces of the alveolar ridge, hard palate, edentulous jaw and alveolar arch in the virtual oral model are enhanced to construct an intraoral alveolar bone model. Finally, the intraoral alveolar bone model is simulated to open and close according to the maximum occlusal gap and the normal occlusal gap using a finite element analysis algorithm.

[0028] Preferably, in step 4, the improved PointNet++ algorithm comprises a set abstraction layer and a feature propagation layer; the set abstraction layer uses the farthest point sampling algorithm to sample the patient's intraoral alveolar bone model as the centroid of the local area of the mouth; and uses the ball query algorithm or the kNN algorithm to obtain a neighborhood point set in the intraoral alveolar bone model according to the centroid coordinates; then uses a multi-layer perception to extract the local features of the local area of the mouth and performs maximum pooling; the feature propagation layer uses an interpolation algorithm to map the high-level features in the local features to the low-level features, thereby obtaining the global features of the intraoral alveolar bone model.

[0029] Preferably, the set abstraction layer comprises a sampling layer, a grouping layer and a PointNet layer; the sampling layer uses an iterative farthest point sampling algorithm to collect a plurality of sampling points in the patient's intraoral alveolar bone model as the centroid of the local area of the mouth;

[0030]

[0031] wherein P new is a new sampling point, argmax represents the iterative farthest point sampling algorithm, p represents a to-be-sampled point, P represents a to-be-sampled point set of the intraoral alveolar bone model, q represents a sampled point, P cur is a sampled point set, and ||·|| represents an Euclidean distance function;

[0032] The grouping layer uses a ball query algorithm to obtain a plurality of neighborhood points in the intraoral alveolar bone model according to the centroid coordinates, to construct a neighborhood point set;

[0033] B(p i ,r)={p j ∈P|||p i -p j ||≤r} (2)

[0034] wherein B() represents the ball query algorithm, p i is the i-th to-be-sampled point, i represents a sampling point serial number, r is a ball radius, B(p i ,r) is a neighborhood point set of p i , and p j represents the j-th to-be-sampled point;

[0035] The PointNet layer converts the original coordinates of the to-be-sampled point into local coordinates relative to the centroid coordinates: x' i =x i -x c ; wherein x' i represents the local coordinates of the to-be-sampled point, x i represents the original coordinates of the i-th to-be-sampled point, and x c represents the centroid coordinates of the local area;

[0036] PointNet layer extracts local features of the local area of the oral cavity by using a multi-layer perceptron (MLP) and performs maximum pooling;

[0037] h(x') = ReLU(W2ReLU(W1x'))

[0038] wherein h() represents a local feature, x' represents a sampling point coordinate in the local area of the oral cavity, W1 and W2 are respectively the first and second weight matrices of the MLP, and ReLU represents an activation function;

[0039] f(B) = maxpool x’∈ Bh(x') (4)

[0040] wherein f() represents a maximum pooling feature of a local area, maxpool represents a maximum pooling layer, and B represents a local area B.

[0041] Preferably, the feature propagation layer realizes the mapping of high-level features to low-level features by using an inverse distance weight function through an interpolation algorithm, thereby obtaining the global feature of the intraoral alveolar bone model.

[0042]

[0043] wherein w i () represents an inverse distance weight value of the i-th neighbor point, x represents a reference point coordinate, d() represents a distance function, s represents a constant, s is 2, and f j () represents an interpolation operation value of the j-th feature, k represents the total number of neighbor points, represents a mapping coefficient of the j-th feature of the i-th neighbor point.

[0044] Preferably, in step S5, the complete denture comprises a base and artificial teeth; the base is tightly attached to the oral mucosa, and is retained by the adsorption force and atmospheric pressure generated by the edge closure, thereby being adsorbed on the upper alveolar ridge and the lower alveolar ridge in the oral cavity; the base and the artificial teeth together restore the facial morphology and function of the patient; the fixed denture comprises a retention body, artificial teeth, and a connecting body; wherein the retention body is fixed on the natural tooth, the bridge restores the morphology and function of the missing tooth, and the connecting body connects the retention body and the artificial teeth.

[0045] Preferably, step S6 comprises the following steps:

[0046] S6-1: The production staff produces the denture model according to the patient's needs, uses a 3D printing cutting device to produce, and simultaneously cuts a hard wax base, thereby combining into a complete denture or a fixed denture.

[0047] S6-2: After the patient wears the complete denture or fixed denture, the data of the feedback of the border position of the base, the fit of the base, the facial ratio, the condylar position, the contraction of the temporal muscle, the centric occlusion, the lateral movement and the protrusion movement are quantified;

[0048] S6-3: The designer corrects the data of the denture model according to the quantified data, and then obtains the final denture data of the denture model.

[0049] Preferably, step S8 comprises the following steps:

[0050] S8-1: The production personnel complete the denture tooth arrangement according to the final denture guide plate by using digital process or traditional process;

[0051] S8-2: The production personnel fix the shape of the denture by using traditional boxing process and silicone rubber or plaster embedding process;

[0052] S8-3: The preliminary denture model completed by boxing and embedding is waxed and waxed, and the mixed denture base polymer is injected into the base model for pressure filling to complete the base production;

[0053] S8-4: The combined denture completed by the base and the denture is polished, polished and cleaned, and the comfort and aesthetics of the combined denture are adjusted;

[0054] S8-5: The combined denture is cast, milled and laser sintered to obtain the final denture model.

[0055] The application improves the PointNet++ algorithm and the improved denture production process, uses digital technology to realize standardized and streamlined denture production, and also establishes a large database to store and analyze the denture shape data of the patient, thereby improving the denture production efficiency and the denture delivery efficiency, reducing the denture processing time, and improving the comfort of the denture. BRIEF DESCRIPTION OF DRAWINGS

[0056] In order to more clearly illustrate the technical solutions in the embodiments of the application or the related art, the following will briefly introduce the drawings needed to be used in the embodiments or related art descriptions. Obviously, the drawings in the following description are only embodiments of the application, and those skilled in the art can obtain other drawings according to the provided drawings without creative labor.

[0057] Figure 1 The principle block diagram of the digital and intelligent denture making method;

[0058] Figure 2 The flowchart of the digital guide plate replication tooth arrangement. DETAILED DESCRIPTION

[0059] The application will be described in detail below in conjunction with the accompanying drawings and examples:

[0060] As shown in Figure 1 and Figure 2 , the digital denture manufacturing method of the application comprises the following steps:

[0061] S1: Obtain and store the anatomical structure data of the patient's oral cavity;

[0062] The anatomical structure data includes two-dimensional data and point cloud data; the point cloud data includes alveolar ridge point cloud data, hard palate point cloud data, edentulous jaw point cloud data, masticatory feature point cloud data and alveolar arch point cloud data; the two-dimensional data includes condyle path inclination, maximum occlusal clearance, normal occlusal clearance and intermaxillary distance;

[0063] S2: Data cleaning is performed on the anatomical structure data, and the AI design software is imported;

[0064] S3: The AI design software generates an intraoral alveolar bone model according to the cleaned anatomical structure data, and performs finite element analysis on the intraoral alveolar bone model;

[0065] S4: In the intraoral alveolar bone model, the denture is arranged according to the anatomical structure data using computer-aided denture arrangement or improved PointNet++ algorithm, and the arranged digital denture is extracted, and then the denture model is obtained, and the denture model is added with a bracket and / or a steel support according to the patient's needs;

[0066] S5: The denture model is used to digitally produce the trial denture using 3D printing cutting equipment to provide the patient with a trial denture; the denture model includes a complete denture model and a fixed denture model;

[0067] S6: The designer corrects the three-dimensional data of the denture model according to the patient's feedback on the trial denture to obtain the design data of the final denture model;

[0068] S7: The AI design software generates final denture guide data according to the design data of the final denture model, and uses 3D printing cutting equipment to manufacture the final denture guide according to the final denture guide data;

[0069] S8: The production staff completes the denture arrangement, shape fixation and detail adjustment according to the final denture guide to determine the final denture model; the detail adjustment includes polishing, polishing and cleaning;

[0070] S9: Use digital process and traditional process to manufacture finished dentures according to the final denture model.

[0071] In the present application, in step S1, the doctor uses an intraoral scanning device to accurately scan the inside of the patient's mouth, or uses a traditional mold taking process to obtain a patient's intraoral model, and then uses a scanner to scan the model to obtain the patient's anatomical structure data; the anatomical structure data of the patient's mouth is stored and analyzed through a large database.

[0072] In the present embodiment, the alveolar ridge is the free edge of the alveolar bone, which is cylindrical in the anterior tooth area and flat in the molar area. The hard palate is composed of the palatine process of the maxillary bone and the horizontal plate of the palatine bone, and the alveolar process is located in the front and side of the palatine process of the maxillary bone. The hard palate is divided into the primary palate (derived from the premaxilla) in the front and the secondary palate (derived from the palatine process of the maxillary process) in the back by the incisive foramen. The surface of the hard palate is covered with mucosa, and the mucosa is closely connected to the bone by a layer of dense connective tissue, which is called the periosteal flap together with the palatal mucosa covering the surface. Edentulous jaws are the complete loss of teeth in the upper and lower jaws; the mastication feature points mainly include the masticatory muscles, teeth, temporomandibular joint, lips, tongue and buccal muscles, etc. which cooperate to complete the mastication movement; the masticatory muscles include the masseter muscle, the temporal muscle, the medial pterygoid muscle and the lateral pterygoid muscle; the alveolar arch refers to the arc-shaped structure formed by the arrangement of teeth in the maxilla and mandible;

[0073] The condylar path inclination mainly describes the angle formed by the movement trajectory of the condyle in the glenoid fossa and the orbito-auricular plane during mandibular protrusive occlusion; the interarch distance is the distance between the top of the upper and lower alveolar ridges when the jaw is in the centric (occlusal) position; the maximum occlusal clearance represents the longest vertical distance between adjacent teeth; the normal occlusal clearance refers to the case that the vertical distance between adjacent teeth is within the normal range; the size of the normal occlusal clearance is 0.5 to 1.5 millimeters;

[0074] In the present application, step S2 includes the following steps:

[0075] S2-1: adaptively divide the point cloud data in the anatomical structure data into a plurality of three-dimensional voxel grids; and denoise the two-dimensional data in the anatomical structure data by using a data cleaning algorithm to obtain denoised two-dimensional data;

[0076] S2-2: calculate the center of mass and the center of the three-dimensional voxel grid to obtain the center of mass voxel block and the center voxel block, respectively;

[0077] S2-3: calculate the spatial domain kernel of the center of mass voxel block according to the point cloud position Euclidean distance by using a Gaussian function, perform Fourier transform on the spatial domain kernel to obtain a spatial domain matrix;

[0078] S2-4: calculate the value domain kernel of the center voxel block according to the gray scale similarity by using a multi-logarithmic function, perform fast Fourier transform on the value domain kernel to obtain a value domain matrix;

[0079] S2-5: multiplying the spatial matrix with the value domain matrix to obtain the denoised point cloud data;

[0080] S2-6: transmitting the denoised point cloud data and the denoised two-dimensional data to the AI design software using a parallel transmission algorithm.

[0081] In the present application, in step S3, the AI design software first correlates the anatomical structure data according to the oral structure using a data correlation algorithm to generate an oral structure data table; then generates a virtual oral model according to the oral structure data table using a three-dimensional reconstruction algorithm; subsequently performs edge enhancement on the virtual oral model using a Gaussian Laplace algorithm, and further enhances the contour traces of the alveolar ridge, hard palate, edentulous jaw, and alveolar arch in the virtual oral model to construct an intraoral alveolar bone model. Finally, the intraoral alveolar bone model is simulated for opening and closing according to the maximum occlusal gap and the normal occlusal gap using a finite element analysis algorithm.

[0082] In the present embodiment, the data correlation algorithm, the three-dimensional reconstruction algorithm, the Gaussian Laplace algorithm, and the finite element analysis algorithm are all commonly used techniques by those skilled in the art, and will not be described in detail here.

[0083] In the present application, in step S4, the improved PointNet++ algorithm includes a set sampling layer and a feature propagation layer; the set sampling layer samples the intraoral alveolar bone model of the patient using a farthest point sampling algorithm as the centroid of the local region of the oral cavity, and uses a ball query algorithm or a kNN algorithm to obtain a neighborhood point set in the intraoral alveolar bone model according to the centroid coordinates, then uses a multi-layer perception to extract local features of the local region of the oral cavity and performs maximum pooling, and the feature propagation layer uses an interpolation algorithm to map high-level features in the local features to low-level features to obtain global features of the intraoral alveolar bone model.

[0084] In the present application, the set abstraction layer includes a sampling layer, a grouping layer, and a PointNet layer; the sampling layer uses an iterative farthest point sampling algorithm to collect a plurality of sampling points in the intraoral alveolar bone model of the patient as the centroid of the local region of the oral cavity;

[0085]

[0086] wherein P new is a new sampling point, argmax represents the iterative farthest point sampling algorithm, p represents a point to be sampled, P represents a set of points to be sampled in the intraoral alveolar bone model, q represents a sampled point, P cur is a set of sampled points, and ||·|| represents the Euclidean distance function;

[0087] The grouping layer uses a ball query algorithm to obtain a plurality of neighborhood points in the intraoral alveolar bone model according to the centroid coordinates to construct a neighborhood point set;

[0088] B(p i ,r)={p j ∈P|||p i -p j ||≤r} (2)

[0089] Where B() represents a ball query algorithm, p i is the i-th sampling point, i represents a sampling point serial number, r is a ball radius, B(p i ,r) is a neighborhood point set of p i , and p j represents the j-th sampling point.

[0090] The PointNet layer converts the original coordinates of the sampling points into local coordinates relative to the centroid coordinates: x' i =x i -x c ; where x' i represents the local coordinates of the sampling points, x i represents the original coordinates of the i-th sampling point, and x c represents the centroid coordinates of the local region.

[0091] The PointNet layer extracts local features of the oral local region using a multi-layer perceptron (MLP) and performs maximum pooling.

[0092] h(x’)=ReLU(W2ReLU(W1x’)) (3)

[0093] Where h() represents the local feature, x' represents the sampling point coordinates in the oral local region, W1 and W2 are the first and second weight matrices of the MLP, respectively, and ReLU represents the activation function.

[0094] f(B)=maxpool x’∈B h(x’) (4)

[0095] Where f() represents the maximum pooling feature of the local region, maxpool represents the maximum pooling layer, and B represents the local region B.

[0096] In the present application, the feature propagation layer realizes the mapping of high-level features to low-level features using an inverse distance weight function through an interpolation algorithm, thereby obtaining the global feature of the intraoral alveolar bone model.

[0097]

[0098] Where w i () represents the inverse distance weight value of the i-th nearest neighbor point, x represents the reference point coordinates, d() represents the distance function, s represents a constant, s is 2, and f j() represents the interpolation operation value of the jth feature, k represents the total number of neighboring points, represents the mapping coefficient of the jth feature of the ith neighboring point.

[0099] In this embodiment, the improved PointNet++ algorithm can capture local and global features of point cloud data at different levels, thereby realizing efficient processing and understanding of point cloud data in the intraoral alveolar bone model.

[0100] First, the key points are selected by the farthest point sampling algorithm, then the neighborhood of each key point is determined by using the ball query algorithm or kNN algorithm, then the neighborhood features are extracted and maximum pooling is performed using a multi-layer perception, and finally the local features are fused into global features by an interpolation algorithm; the fine features of the local region in the point cloud can be extracted; through the feature propagation layer, the improved PointNet++ algorithm can understand the structure of the entire point cloud from the local features; through sampling and local processing, the computational amount is reduced and the processing efficiency is improved.

[0101] In the present application, in step S5, the complete denture comprises a base and artificial teeth; the base is tightly attached to the oral mucosa, and is retained by the adsorption force and atmospheric pressure generated by the edge closure, and is further adsorbed on the upper alveolar ridge and the lower alveolar ridge in the oral cavity; the base and the artificial teeth together restore the facial morphology and function of the patient; the fixed denture comprises a retention body, artificial teeth and a connecting body; wherein the retention body is fixed on the natural tooth, the bridge restores the morphology and function of the missing tooth, and the connecting body connects the retention body and the artificial tooth.

[0102] In the present application, in step S6, the following steps are included:

[0103] S6-1: The production staff produces the denture model according to the patient's needs, uses the 3D printing cutting equipment to produce, and simultaneously cuts the hard wax base, and then combines the complete denture or fixed denture;

[0104] S6-2: According to the feedback of the occlusion of the base edge position, the base fit, the facial ratio, the condylar position, the temporal muscle contraction activity, the centric occlusion, the lateral movement and the protrusion movement after the patient wears the complete denture or fixed denture, the data is quantified;

[0105] S6-3: The designer corrects the data of the denture model according to the quantified data, and then obtains the final denture data of the denture model.

[0106] In the present application, in step S8, the following steps are included:

[0107] S8-1: The production staff arranges the teeth of the denture according to the final denture guide using digital process or traditional process;

[0108] S8-2: The production staff uses a traditional boxing process and a silicone rubber or plaster embedding process to fix the denture shape;

[0109] S8-3: The preliminary denture model that has completed boxing and embedding is subjected to wax boiling and wax injection, and the mixed denture base polymer is injected into the base model for pressure filling to complete the base production;

[0110] S8-4: The combined denture that has completed the base and denture installation is polished, polished and cleaned, and the comfort and aesthetics of the combined denture are adjusted;

[0111] S8-5: The combined denture is cast, milled and laser sintered to obtain the final denture model.

[0112] Embodiment:

[0113] The doctor uses an oral scanning device to accurately scan the inside of the patient's mouth or uses a traditional mold taking process to obtain the patient's oral cavity model, and uses an oral scanning device to scan the patient's oral cavity model data to obtain the patient's oral cavity anatomical structure data; The anatomical structure data of the patient's oral cavity is stored and analyzed by a large database; The anatomical structure data includes two-dimensional data and point cloud data; The point cloud data includes alveolar ridge point cloud data, hard palate point cloud data, edentulous jaw point cloud data, chewing feature point cloud data and alveolar arch point cloud data; The two-dimensional data includes condyle path slope, maximum occlusal clearance, normal occlusal clearance and intermaxillary distance;

[0114] The point cloud data in the anatomical structure data is adaptively voxelized to form a plurality of three-dimensional voxel grids; The two-dimensional data in the anatomical structure data is denoised using a data cleaning algorithm to obtain denoised two-dimensional data; The centroid and center of the three-dimensional voxel grid are calculated to obtain the centroid voxel block and the center voxel block, respectively; The spatial domain kernel of the centroid voxel block is calculated using a Gaussian function according to the Euclidean distance of the point cloud position, and the spatial domain kernel is subjected to Fourier transform to obtain a spatial matrix; The value domain kernel of the center voxel block is calculated using a multi-logarithmic function according to the gray similarity, and the value domain kernel is subjected to fast Fourier transform to obtain a value domain matrix; The denoised point cloud data is obtained by multiplying the spatial matrix and the value domain matrix; The denoised point cloud data and the denoised two-dimensional data are transmitted to an AI design software using a parallel transmission algorithm;

[0115] The AI design software firstly utilizes a data association algorithm to associate the anatomical structure data according to the oral structure to generate an oral structure data table; then utilizes a three-dimensional reconstruction algorithm to generate a virtual oral model according to the oral structure data table; subsequently utilizes a Gaussian Laplace algorithm to perform edge enhancement on the virtual oral model, and further enhances the contour traces of the alveolar ridge, hard palate, edentulous jaw and alveolar arch in the virtual oral model to construct an intraoral alveolar bone model; finally utilizes a finite element analysis algorithm to simulate the opening and closing of the intraoral alveolar bone model according to the maximum occlusal gap and the normal occlusal gap;

[0116] In the intraoral alveolar bone model, a computer-aided tooth arrangement method or an improved PointNet++ algorithm is used to arrange the denture according to the anatomical structure data, and the arranged digital denture is extracted to obtain a denture model; and according to the patient's needs, a support and / or steel support is added to the denture model; the improved PointNet++ algorithm includes a set sampling layer and a feature propagation layer; the set sampling layer uses the farthest point sampling algorithm to sample the patient's intraoral alveolar bone model as the centroid of the local area of the oral cavity; and the centroid coordinates are used to obtain a neighborhood point set in the intraoral alveolar bone model using a ball query algorithm or a kNN algorithm; then the local features of the local area of the oral cavity are extracted using a multi-layer perception, and maximum pooling is performed; the feature propagation layer uses an interpolation algorithm to map high-level features in the local features to low-level features, thereby obtaining global features of the intraoral alveolar bone model;

[0117] The denture model is used for digital production of the trial denture by using a 3D printing cutting equipment to provide the patient with a trial denture; the denture model includes a complete denture and a fixed denture; the complete denture includes a base and a bridge; the base is tightly attached to the oral mucosa, and is retained by the adsorption force and atmospheric pressure generated by the edge closure, thereby being adsorbed on the upper and lower alveolar ridges of the oral cavity to restore the patient's facial form and function; the fixed denture includes a retention body, a bridge and a connecting body; the retention body is fixed to the natural tooth, the bridge restores the form and function of the missing tooth, and the connecting body connects the retention body and the bridge; the bridge is an artificial tooth; the production personnel produce the denture model using the 3D printing cutting equipment according to the patient's needs, and simultaneously cut a hard wax support to combine the complete denture or the fixed denture; the feedback of the base edge, the base fit, the facial proportion, the condylar position, the temporal muscle contraction activity, the centric occlusion, the lateral movement and the protrusive movement after the patient wears the complete denture or the fixed denture is quantified; the designer corrects the data of the denture model according to the quantified data to obtain final denture data; the AI design software generates final denture guide data according to the final denture data, and the production personnel produce the final denture guide according to the final denture guide data using the 3D printing cutting equipment;

[0118] The production staff completes denture tooth arrangement according to the final denture guide plate by using a digital process or a traditional process; the production staff fixes the shape of the denture by using a traditional boxing process and a silicone rubber embedding process or a plaster embedding process; the production staff performs wax boiling and wax injection on the preliminary denture model after boxing and embedding, and pours the mixed denture base polymer into the base model for pressure boiling and glue boiling to complete the production of the base; the production staff polishes, polishes and cleans the combined denture after completing the base and the denture installation, adjusts the comfort and aesthetics of the combined denture; the production staff casts, mills and laser sintering to obtain the final denture model; finally, the production staff obtains the finished denture according to the final denture model by using a digital process and a traditional process.

Claims

1. A method for manufacturing digital dentures, characterized in that: Includes the following steps: S1: Acquire and store anatomical data of the patient's oral cavity; Anatomical data includes two-dimensional data and point cloud data; the point cloud data includes alveolar ridge point cloud data, hard palate point cloud data, edentulous jaw point cloud data, masticatory feature point cloud data, and alveolar arch point cloud data; the two-dimensional data includes condylar inclination, maximum occlusal space, normal occlusal space, and intermaxillary distance. S2: Clean the anatomical structure data and import it into AI design software; S3: The AI ​​design software generates an intraoral alveolar bone model based on the cleaned anatomical structure data, and performs finite element analysis on the intraoral alveolar bone model. S4: Using computer-aided tooth arrangement or the improved PointNet++ algorithm in the intraoral alveolar bone model, the prosthesis is arranged according to the anatomical structure data, and the digital prosthesis with the arrangement is extracted to obtain the prosthesis model. Then, the framework and / or metal frame are added to the prosthesis model with one click according to the patient's needs. S5: Digitally produce denture models for trial fitting using 3D printing cutting equipment, so as to provide them to patients for denture trial fitting; denture models include complete denture models and fixed denture models; S6: The designer corrects the three-dimensional data of the denture model based on the patient's feedback on the denture trial, and obtains the design data of the final denture model; S7: AI design software generates final denture guide plate data based on the design data of the final denture model, and uses 3D printing cutting equipment to manufacture the final denture guide plate based on the final denture guide plate data; S8: The production staff completes tooth arrangement, shape fixation, and detail adjustments based on the final denture guide plate to determine the final denture model; detail adjustments include grinding, polishing, and cleaning; S9: Using digital and traditional processes, produce finished dentures based on the final denture model; In step S4, the improved PointNet++ algorithm includes an ensemble sampling layer and a feature propagation layer. The ensemble sampling layer uses the farthest point sampling algorithm to sample the patient's intraoral alveolar bone model as the centroid of the local oral region. It then uses the ball query algorithm or kNN algorithm to obtain the set of neighborhood points in the intraoral alveolar bone model based on the centroid coordinates. Next, it uses a multilayer perceptron to extract the local features of the local oral region and performs max pooling. The feature propagation layer uses an interpolation algorithm to map high-level features in the local features to low-level features, thereby obtaining the global features of the intraoral alveolar bone model. The ensemble sampling layer includes a sampling layer, a grouping layer, and a PointNet layer; the sampling layer uses an iterative farthest point sampling algorithm to collect several sampling points in the patient's intraoral alveolar bone model as the centroid of the local oral region; （1） in, For the new sampling point, This represents the iterative farthest point sampling algorithm. Indicates the sampling point. This represents the set of sampling points in the intraoral alveolar bone model. Indicates the points that have been sampled. It is the set of sampled points. Represents the Euclidean distance function; The grouping layer uses the ball query algorithm to obtain several neighborhood points in the intraoral alveolar bone model based on the centroid coordinates, and constructs a neighborhood point set; （2） in, Represents the ball query algorithm, It is the first One sampling point, Indicates the sampling point sequence number. It is the radius of the sphere. yes The set of neighborhood points, Indicates the first One sampling point; The PointNet layer converts the original coordinates of the points to be sampled into local coordinates relative to the centroid coordinates: ;in, Indicates the local coordinates of the point to be sampled. Indicates the first The original coordinates of the points to be sampled. Indicates the centroid coordinates of a local region; The PointNet layer uses a multilayer perceptron to extract local features of the oral cavity and performs max pooling. （3） in, Indicates local features, This indicates the coordinates of the sampling point within a localized area of ​​the oral cavity. and These are the first and second weight matrices of the MLP, respectively. Indicates the activation function; （4） in: This represents the max pooling feature of a local region. Indicates the max pooling layer. This represents the local region B.

2. The method for manufacturing digital dentures according to claim 1, characterized in that: In step S1, the doctor uses an intraoral scanning device to perform a precise scan of the patient's oral cavity, or uses a traditional impression-taking process to obtain an intraoral model of the patient, and then uses a scanner to perform an external scan of the model to obtain the anatomical structure data of the patient's oral cavity; the anatomical structure data of the patient's oral cavity is stored and analyzed through a large database.

3. The digital denture fabrication method according to claim 1, characterized in that: Step S2 includes the following steps: S2-1: Adaptively divide the point cloud data in the anatomical structure data into several three-dimensional voxel grids; use a data cleaning algorithm to denoise the two-dimensional data in the anatomical structure data to obtain denoised two-dimensional data. S2-2: Calculate the centroid and center of the 3D voxel grid, and obtain the centroid voxel block and the center voxel block respectively; S2-3: Calculate the spatial domain kernel of the centroid voxel block based on the Euclidean distance of the point cloud position using the Gaussian function, and perform a Fourier transform on the spatial domain kernel to obtain the spatial domain matrix. S2-4: Calculate the value range kernel of the central voxel block based on gray-level similarity using the multiple logarithmic function, and perform a fast Fourier transform on the value range kernel to obtain the value range matrix; S2-5: Multiply the spatial domain matrix with the range matrix to obtain the denoised point cloud data; S2-6: The denoised point cloud data and denoised 2D data are transmitted to the AI ​​design software using a parallel transmission algorithm.

4. The method for manufacturing digital dentures according to claim 1, characterized in that: In step S3, the AI ​​design software first uses a data association algorithm to associate anatomical structure data based on the oral cavity structure to generate an oral cavity structure data table; then, it uses a 3D reconstruction algorithm to generate a virtual oral cavity model based on the oral cavity structure data table; subsequently, it uses the Gaussian Laplace algorithm to enhance the edges of the virtual oral cavity model, thereby enhancing the contour traces of the alveolar ridge, hard palate, edentulous jaw, and alveolar arch in the virtual oral cavity model, and constructing an intraoral alveolar bone model; finally, it uses a finite element analysis algorithm to perform opening and closing simulations on the intraoral alveolar bone model based on the maximum occlusal gap and normal occlusal gap.

5. The method for manufacturing digital dentures according to claim 1, characterized in that: The feature propagation layer uses an interpolation algorithm and an inverse distance weight function to map high-level features to low-level features, thereby obtaining the global features of the intraoral alveolar bone model. （5） （6） in, Indicates the first The inverse distance weights of the nearest neighbors. Indicates the coordinates of the reference point. Represents the distance function. Represents a constant. Take 2, Indicates the first The interpolated values ​​of each feature. This represents the total number of nearest neighbors. Indicates the first The nearest neighbor point The mapping coefficients of each feature.

6. The method for manufacturing digital dentures according to claim 1, characterized in that: In step S5, the complete denture includes a base and artificial teeth; the base fits tightly to the oral mucosa and is retained by the suction force generated by the edge closure and atmospheric pressure, and then adheres to the upper and lower alveolar ridges in the oral cavity; the base and artificial teeth together restore the patient's facial shape and function; the fixed denture includes a retainer, artificial teeth and connectors; the retainer is fixed to the natural teeth, the bridge restores the shape and function of the missing teeth, and the connectors connect the retainer and artificial teeth.

7. The method for manufacturing digital dentures according to claim 6, characterized in that: Step S6 includes the following steps: S6-1: The production staff uses 3D printing cutting equipment to produce the denture model according to the patient's needs, and simultaneously cuts the hard wax base, and then assembles it into a complete denture or fixed denture. S6-2: Based on the feedback of patients after wearing complete or fixed dentures, data on the position of the denture base edge, denture fit, facial proportion, condylar position, temporalis muscle contraction movement, centric occlusion, lateral movement and protrusion movement are quantified. S6-3: The designer corrects the denture model based on the quantitative data, and then obtains the final denture model data.

8. The method for manufacturing digital dentures according to claim 1, characterized in that: Step S8 includes the following steps: S8-1: Production staff complete the denture tooth arrangement using digital or traditional processes based on the final denture guide plate; S8-2: Production staff use traditional boxing processes and silicone rubber or plaster embedding processes to fix the shape of the denture; S8-3: The preliminary denture model that has been boxed and embedded is boiled and rinsed with wax, and the mixed denture base polymer is injected into the base model for pressure filling to complete the base fabrication; S8-4: Grind, polish, and clean the assembled denture after the base and denture are installed, and adjust the comfort and aesthetics of the assembled denture. S8-5: Casting, milling and laser sintering of the composite denture to obtain the final denture model.

Images