High-precision digital complete denture 3D printing method

Abstract

The invention relates to the technical field of denture manufacturing, and discloses a high-precision digital complete denture 3D printing method which comprises the following steps: obtaining an initial denture model of a target object; acquiring occlusal point adjustment data; acquiring actual adjustment data; adjusting the initial false tooth model based on the actual adjustment data, and converting the adjusted initial false tooth model into a 3D printing instruction for 3D printing; according to the method, stress distribution during occlusion is simulated through finite element analysis, a three-dimensional occlusion thermodynamic diagram is obtained and input into a constructed CNN network model, occlusion point adjustment data are obtained, adjustment data of effective occlusion contact points are mapped into a solid model, and actual adjustment data are obtained; three-dimensional data of the oral cavity of a patient is obtained through a scanner, digital analysis is conducted in combination with a neural network model, the denture manufacturing precision is improved, manufacturing is conducted in a 3D printing mode, the manufacturing efficiency is high, the patient feels comfortable when wearing the denture and occludes the denture well, and the problem of diet difficulty of the implant patient is solved in time.

Description

Technical Field

[0001] The present invention relates to the technical field of denture production, and more specifically to a high-precision digital full-mouth denture 3D printing method. Background Art

[0002] Many elderly people have lost their teeth due to periodontal disease, caries and other reasons, and need suitable dentures to restore their chewing, aesthetic and pronunciation functions. Traditional customized removable dentures are made of denture polymers and synthetic resin teeth through embedding injection molding, and are made on-site based on the impression of the patient's oral implants, so as to provide patients with personalized and immediate restoration solutions for the upper part of the implant to meet the needs of patients with missing implants.

[0003] However, traditional denture production requires on-site production based on impressions of the patient's oral implants, which has high requirements on personnel skills and is purely manual. Any carelessness in any link will lead to abnormal positioning or occlusion of the denture, and fail to achieve the expected repair effect. The accuracy is poor and the processing takes a long time. In view of this, the present invention proposes a high-precision digital full-mouth denture 3D printing method, which obtains the patient's oral three-dimensional data through a scanner, and combines it with a neural network model for digital analysis to improve the accuracy of denture production, and adopts 3D printing to produce it. Summary of the invention

[0004] In order to overcome the above-mentioned defects of the prior art, the present invention provides a high-precision digital full denture 3D printing method to solve the problems existing in the above-mentioned background technology.

[0005] The present invention provides the following technical solution: a high-precision digital full denture 3D printing method, comprising the following steps:

[0006] Step S1: Acquire the oral three-dimensional data of the target object, and obtain the initial denture model of the target object based on the constructed knowledge graph matching; the target object is the patient who needs full denture production;

[0007] Step S2: Based on the oral three-dimensional data and the corresponding initial denture model, finite element analysis is used to simulate the stress distribution during occlusion, a three-dimensional occlusal heat map is obtained, and the occlusal heat map is input into the constructed CNN network model to obtain the occlusal point adjustment data;

[0008] Step S3: obtaining the confidence of the occlusal point adjustment data, performing confidence judgment to obtain the effective occlusal contact points, and mapping the adjustment data of the effective occlusal contact points to the physical model to obtain the actual adjustment data;

[0009] Step S4: The initial denture model is adjusted based on the actual adjustment data, and converted into a 3D printing instruction for 3D printing.

[0010] Preferably, the oral three-dimensional data in step S1 includes jaw relationship and oral tissue data, and the jaw relationship includes vertical jaw relationship and horizontal jaw relationship; the vertical jaw relationship is the vertical distance, and the determination of the horizontal jaw relationship is the determination of the central relationship.

[0011] Preferably, the constructed knowledge graph is an oral knowledge graph, including entities, entity attribute information, relationships between entities, and entity types. The entity attribute information is a label describing the entity attributes; the relationship between entities represents the connection between different entities, describing various associations and interactions between entities in the knowledge graph; the entity type defines the category to which the entity belongs, and defines the common attributes and relationships between entities; the oral knowledge graph is constructed in the following manner:

[0012] The oral 3D data of n edentulous patients are collected in advance, and all of the n edentulous patients need to make full dentures. The 3D models of the finished dentures of the n edentulous patients are obtained, and the oral 3D data of each edentulous patient are formed into a data set, that is, there are n data sets, each data set is an entity in the knowledge graph, and the entity attribute information is the basic information of the edentulous patient corresponding to the entity, and the basic information includes but is not limited to age, gender, name, etc. The entity type is the 3D model of the finished denture of the edentulous patient corresponding to the entity.

[0013] Preferably, the initial denture model is obtained by:

[0014] Match the oral three-dimensional data of the target object with the entities in the oral knowledge graph, take the entity with the highest matching degree as the successfully matched entity, mark it as the matching entity, and obtain the entity type corresponding to the matching entity as the initial denture model of the target object; that is, the three-dimensional model of the finished denture of the edentulous patient corresponding to the matching entity is used as the initial denture model of the target object;

[0015] The method of matching the target object's oral 3D data with the entities in the oral knowledge graph is as follows:

[0016] The oral three-dimensional data of the target object is formed into a target data set, and the target data set and the data set corresponding to the entities in the oral knowledge graph are preprocessed, and matching is performed by extracting the feature vectors of the data in the data set.

[0017] Preferably, the three-dimensional occlusal heat map is an RGB image, in which the red channel represents the contact pressure of the anterior teeth area, the green channel represents the contact pressure of the left posterior teeth area, and the blue channel represents the contact pressure of the right posterior teeth area;

[0018] Obtain occlusal contact points based on the three-dimensional occlusal heat map and mark them as 1, 2, 3, ..., N; input the three-dimensional occlusal heat map into the trained CNN network model, map the N occlusal contact points to the corresponding pixel positions of the output tensor, and output the three-dimensional adjustment amount of each pixel point position;

[0019] The CNN network model mainly includes an input layer, a convolution layer, a pooling layer, a comprehensive loss function, a fully connected layer and an output layer; the input layer is used to input data, namely, a three-dimensional occlusion heat map; the convolution layer is used to extract primary and advanced features of the data, and the comprehensive loss function combines the position error term, the contact area ratio error term and the smoothing loss; the output layer is used to output data, namely, the three-dimensional adjustment amount of the occlusal contact point and the predicted occlusal contact area ratio.

[0020] Preferably, the specific method of training the CNN network model is:

[0021] Based on the pre-collected oral 3D data and 3D occlusal heat map of n edentulous patients, the corresponding denture production data are obtained, and the oral 3D data of each edentulous patient and the corresponding denture production data are used as a group of research data, so there are n groups of research data, which are divided into a training set, a validation set and a test set;

[0022] Each set of training data is used as the input of the CNN network model, and the predicted bite point adjustment data is used as the output, with the training goal of minimizing the sum of the prediction errors of all training data; the calculation formula of the prediction error is: Where Z k is the prediction error, k is the group number corresponding to the research data, a ki is the occlusal point adjustment data predicted for the i-th occlusal contact point in the k-th group of research data, b ki is the actual occlusal point adjustment data of the i-th occlusal contact point in the k-th group of research data, that is, the occlusal point adjustment data existing in the denture production data; the CNN network model is trained until the sum of the prediction errors reaches convergence, and the training is stopped to obtain a trained CNN network model.

[0023] Preferably, the formula of the comprehensive loss function is expressed as:

[0024] Loss total =α·L weizhi +β·L ratio +γ·L smooth Among them, Loss total represents the comprehensive loss function, L weizhi Represents the position error term loss, L ratio Represents the contact area ratio error term loss, L smoothrepresents smoothing loss; α, β, and γ are the corresponding weight coefficients respectively;

[0025] in, represents the predicted value of the three-dimensional displacement vector of the i-th occlusal contact point, represents the actual value of the three-dimensional displacement vector of the i-th occlusal contact point; i = 1, 2, 3, ..., N; N is the total number of contact points;

[0026] in, represents the predicted occlusal contact area ratio, Indicates the actual occlusal area contact ratio.

[0027] Preferably, the bite point adjustment data is the three-dimensional position adjustment amount of the pixel point corresponding to the bite contact point, the horizontal adjustment amount is marked as Δx, the longitudinal adjustment amount is marked as Δy, and the vertical adjustment amount is marked as Δz, and the adjustment direction can be represented by a positive or negative sign;

[0028] The confidence of the occlusal point adjustment data is obtained, that is, the confidence of the pixel corresponding to the occlusal contact point is obtained, and whether the pixel is a valid pixel is determined. If it is a valid pixel, the occlusal contact point corresponding to the pixel is a valid occlusal contact point.

[0029] Preferably, in step S3, the adjustment data of the effective occlusal contact points are mapped to the physical model, and the specific method of obtaining the actual adjustment data is:

[0030] Convert the adjustment amount of the effective occlusal contact point into a physical size;

[0031] The adjustment data converted into physical dimensions is mapped to a solid model of the target object to obtain actual adjustment data, wherein the solid model is the actual three-dimensional occlusal plane.

[0032] Preferably, the formula for converting the adjustment amount of the effective occlusal contact point into a physical size is expressed as:

[0033]

[0034] Where Δx wl , Δy wl , Δz wl Represents the physical size adjustment in the horizontal, longitudinal and vertical directions, Δx xs , Δy xs , Δx xs Respectively represent the horizontal, longitudinal and vertical pixel adjustment amounts, W x Indicates the actual width of the horizontal occlusal surface, W y Indicates the actual longitudinal height of the occlusal surface; W zIndicates the actual vertical depth of the occlusal surface, Pit x Indicates the pixel ratio value corresponding to the actual width of the occlusal surface, Pit y Indicates the pixel ratio value corresponding to the actual height of the occlusal surface, Pit z Indicates the pixel ratio value corresponding to the actual depth of the occlusal surface.

[0035] Technical effects and advantages of the present invention:

[0036] The present invention is provided with step S2 and step S3, which is conducive to simulating the stress distribution during occlusion by using finite element analysis, obtaining a three-dimensional occlusal heat map, and inputting it into the constructed CNN network model to obtain the occlusal point adjustment data, and mapping the adjustment data of the effective occlusal contact point to the physical model to obtain the actual adjustment data; the three-dimensional data of the patient's oral cavity is obtained by a scanner, and digital analysis is performed in combination with the neural network model to improve the accuracy of denture production, and 3D printing is used for production, with high production efficiency, high precision, patient comfort and good occlusion, and timely resolution of the dietary difficulties of implant patients. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 This is a flow chart of the high-precision digital full denture 3D printing method of the present invention. DETAILED DESCRIPTION

[0038] The technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the present invention. In addition, the forms of the various structures described in the following embodiments are merely illustrative. The high-precision digital full denture 3D printing method involved in the present invention is not limited to the various structures described in the following embodiments. All other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0039] like Figure 1 As shown, the present invention provides a high-precision digital full-mouth denture 3D printing method, comprising the following steps:

[0040] Step S1: Acquire the oral three-dimensional data of the target object, and obtain the initial denture model of the target object based on the constructed knowledge graph matching; the target object is the patient who needs full denture production;

[0041] Step S2: Based on the oral three-dimensional data and the corresponding initial denture model, finite element analysis is used to simulate the stress distribution during occlusion, a three-dimensional occlusal heat map is obtained, and the occlusal heat map is input into the constructed CNN network model to obtain the occlusal point adjustment data;

[0042] Step S3: obtaining the confidence of the occlusal point adjustment data, performing confidence judgment to obtain the effective occlusal contact points, and mapping the adjustment data of the effective occlusal contact points to the physical model to obtain the actual adjustment data;

[0043] Step S4: The initial denture model is adjusted based on the actual adjustment data, and converted into a 3D printing instruction for 3D printing.

[0044] In this embodiment, it should be specifically explained that the oral three-dimensional data in step S1 includes but is not limited to jaw position relationship and oral tissue data, and the oral tissue data includes but is not limited to alveolar bone density, alveolar bone morphology, soft tissue morphology, and the positions of key anatomical structures such as the maxillary sinus, nasal cavity, and mandibular nerve canal, etc., which can be obtained through cone beam computed tomography. The soft tissue morphology, i.e. the morphology of soft tissues such as gums and lips and cheeks, can be obtained through an intraoral scanner;

[0045] The jaw position relationship includes the vertical jaw position relationship and the horizontal jaw position relationship; the vertical jaw position relationship is the vertical distance, which is the distance from the nose bottom to the chin bottom when the natural dentition is in the middle jaw, that is, the distance of the lower third of the face. The vertical distance can be obtained by the rest jaw method, measuring the natural relaxed distance from the nose bottom to the chin bottom, deducting 2-3mm rest gap as the vertical reference value, and the formula is expressed as: H 垂直 =H 息止 -Δh; where H 垂直 Indicates the vertical distance, H 息止 It represents the resting distance, that is, the distance from the bottom of the nose to the bottom of the chin when naturally relaxed, and Δh represents the resting interval;

[0046] The determination of the horizontal jaw relationship is to determine the centric relationship. The centric relationship refers to the mandibular condyle being located in the physiological posterior position in the center of the articular concavity without restriction. Only in this position does the patient feel that the temporomandibular joint is not tense and is comfortable, and the chewing muscles are strong and the chewing efficiency is high. The acquisition method is: based on the three-part ratio of the hairline-brow point-nasion point-chin base, ensure that the jaw midline is aligned with the facial midline, mark the centric relationship position with the silicone bite recording material, and record the physiological posterior position of the mandibular condyle.

[0047] In this embodiment, it should be specifically explained that the constructed knowledge graph is an oral knowledge graph, including entities, entity attribute information, relationships between entities, and entity types. The entity attribute information is a label describing the entity attributes; the relationship between entities represents the connection between different entities, describing various associations and interactions between entities in the knowledge graph. The entity type defines the category to which the entity belongs, and defines the common attributes and relationships between entities. The oral knowledge graph is constructed in the following manner:

[0048] Collect oral 3D data of n edentulous patients in advance, all of whom need to make full dentures, obtain 3D models of finished dentures of the n edentulous patients, and form a data set with the oral 3D data of each edentulous patient, that is, there are n data sets, each data set is used as an entity in the knowledge graph, and the entity attribute information is the basic information of the edentulous patient corresponding to the entity, and the basic information includes but is not limited to age, gender, name, etc., and the entity type is the 3D model of the finished denture of the edentulous patient corresponding to the entity;

[0049] Match the oral three-dimensional data of the target object with the entities in the oral knowledge graph, take the entity with the highest matching degree as the successfully matched entity, mark it as the matching entity, and obtain the entity type corresponding to the matching entity as the initial denture model of the target object; that is, the three-dimensional model of the finished denture of the edentulous patient corresponding to the matching entity is used as the initial denture model of the target object;

[0050] The method of matching the target object's oral 3D data with the entities in the oral knowledge graph is as follows:

[0051] The oral three-dimensional data of the target object is formed into a target data set, and the target data set and the data set corresponding to the entities in the oral knowledge graph are preprocessed to ensure the uniformity of the data, and matching is performed by extracting the feature vectors of the data in the data set.

[0052] In this embodiment, it should be specifically explained that the three-dimensional occlusal heat map is an RGB image, in which the red channel represents the contact pressure of the anterior teeth area, the green channel represents the contact pressure of the left posterior teeth area, and the blue channel represents the contact pressure of the right posterior teeth area; the three-dimensional occlusal heat map intuitively displays the distribution of contact pressure during occlusion through color gradient, and its purpose is to convert the mechanical distribution of the three-dimensional occlusal surface into two-dimensional image features, so as to facilitate CNN to extract spatial patterns;

[0053] Obtain occlusal contact points based on the three-dimensional occlusal heat map and mark them as 1, 2, 3, ..., N; input the three-dimensional occlusal heat map into the trained CNN network model, map the N occlusal contact points to the corresponding pixel positions of the output tensor, and output the three-dimensional adjustment amount of each pixel point position;

[0054] The CNN network model mainly includes an input layer, a convolution layer, a pooling layer, a comprehensive loss function, a fully connected layer and an output layer; the input layer is used to input data, namely, a three-dimensional occlusal heat map; the convolution layer is used to extract primary and advanced features of the data, the pooling layer performs average pooling or maximum pooling, and the comprehensive loss function combines the position error term, the contact area ratio error term and the smoothing loss to improve the robustness of the model while improving the accuracy of the occlusal contact point adjustment amount; the fully connected layer adopts L2 regularization to prevent overfitting, and the output layer is used to output data, namely, the three-dimensional adjustment amount of the occlusal contact point and the predicted occlusal contact area ratio;

[0055] The specific method of training the CNN network model is:

[0056] Based on the pre-collected oral three-dimensional data and three-dimensional occlusal heat map of n edentulous patients, the corresponding denture production data are obtained, and the oral three-dimensional data of each edentulous patient and its corresponding denture production data are used as a group of research data, then there are n groups of research data, and the n groups of research data are divided into a training set, a validation set and a test set; the denture production data is the adjustment data in the denture production process, such as the cutting data and grinding data of the resin disc in the denture design process; it reflects the adjustment of the wearing conditions of different patients in the denture production process. When the denture is produced, a digital model is first formed according to the oral three-dimensional data, and the computer-aided design technology is applied to edit and design the oral three-dimensional data to form a trial model, and then the final denture model is adjusted according to the actual trial wearing situation. Therefore, there is occlusal point adjustment data in the denture production data; the division ratio of the training set, validation set and test set can be set by the personnel in this field. In this embodiment, the division ratio is selected as 70%, 20% and 10%;

[0057] Each set of training data is used as the input of the CNN network model, and the predicted bite point adjustment data is used as the output, with the training goal of minimizing the sum of the prediction errors of all training data; the calculation formula of the prediction error is: Where Z k is the prediction error, k is the group number corresponding to the research data, a ki is the occlusal point adjustment data predicted for the i-th occlusal contact point in the k-th group of research data, b ki is the actual occlusal point adjustment data of the i-th occlusal contact point in the k-th group of research data, that is, the occlusal point adjustment data existing in the denture manufacturing data; the CNN network model is trained until the sum of the prediction errors reaches convergence, and the training is stopped to obtain a trained CNN network model;

[0058] The validation set is input into the trained model to select the hyperparameters in the model, and the test set is used to evaluate and optimize the model to complete the construction of the CNN network model.

[0059] In this embodiment, it should be specifically explained that the formula of the comprehensive loss function is expressed as:

[0060] Loss total =α·L weizhi +β·L ratio +γ·L smooth Among them, Loss total Represents the comprehensive loss function, L weishi Represents the position error term loss, L ratio Represents the contact area ratio error term loss, L zmooth represents smoothing loss; α, β, and γ are corresponding weight coefficients respectively; the specific values ​​can be set or modified by those skilled in the art according to actual conditions; in this embodiment, α=0.5, β=0.4, and γ=0.1 are selected;

[0061] in, represents the predicted value of the three-dimensional displacement vector of the i-th occlusal contact point, represents the actual value of the three-dimensional displacement vector of the i-th occlusal contact point; i = 1, 2, 3, ..., N; N is the total number of contact points;

[0062] in, represents the predicted occlusal contact area ratio, represents the actual occlusal area contact ratio; the occlusal contact area ratio is calculated as follows: the area of ​​each occlusal contact point is accumulated to obtain the total contact area, and the total contact area is divided by the theoretical maximum contact area to obtain the occlusal contact area ratio; the occlusal contact area ratio refers to the ratio of the actual contact area to the theoretical maximum contact area during the occlusion process, and this ratio reflects the tightness and stability of the occlusion;

[0063] L smooth Used to adjust the model to make the calculation results closer to the real physical phenomenon, reduce the impact of sudden or abnormal values ​​by weighted averaging of actual data and predicted data, and improve the stability and robustness of the model;

[0064] The selection of the comprehensive loss function combines the position error of the occlusal contact point and the contact area ratio error. For the CNN network model, the position of the occlusal contact point can be better optimized in predicting the adjustment data of the occlusal contact point, so that the denture after adjusting the occlusal contact point is more helpful for the bite of edentulous patients, more comfortable to wear, and more practical.

[0065] In this embodiment, it is necessary to specifically explain that the occlusal point adjustment data, i.e., the three-dimensional position adjustment amount of the pixel point corresponding to the occlusal contact point, is marked as Δx for the horizontal adjustment amount, Δy for the longitudinal adjustment amount, and Δz for the vertical adjustment amount, and the adjustment direction can be represented by positive and negative signs; in the oral three-dimensional model, the horizontal direction refers to the direction parallel to the ground, i.e., the direction parallel to the dental arch, which can be used to describe the width of the dentition or the horizontal distance between the teeth; the longitudinal direction refers to the direction from the crown to the root of the tooth, or from the incisal end to the neck of the tooth, which can be used to describe the height of the tooth in the oral cavity or the axial direction of the tooth; the vertical direction refers to the direction perpendicular to the ground, i.e., the direction perpendicular to the dental arch, which can be used to describe the depth of the tooth in the oral cavity or the vertical distance between the teeth, such as the depth of the alveolar bone or the implantation depth of the tooth in the alveolar bone;

[0066] Obtain the confidence of the occlusal point adjustment data, that is, obtain the confidence of the pixel corresponding to the occlusal contact point, and judge whether the pixel is a valid pixel. If it is a valid pixel, the occlusal contact point corresponding to the pixel is a valid occlusal contact point. The formula for confidence judgment is expressed as: valid pixel = {(c, d) | confidence cd ≥YU}; wherein, (c, d) represents the pixel point (c, d), YU represents the confidence threshold, and in this embodiment, YU=0.8; the confidence can be obtained by setting the confidence channel of CNN. If there is a pixel point (c, d) whose confidence is greater than or equal to the confidence threshold, the pixel point (c, d) is judged to be a valid pixel point, and the occlusal contact point corresponding to (c, d) is a valid occlusal contact point.

[0067] In this embodiment, it should be specifically explained that in step S3, the adjustment data of the effective occlusal contact point is mapped to the physical model, and the specific method of obtaining the actual adjustment data is as follows:

[0068] Convert the adjustment amount of the effective occlusal contact point into a physical size;

[0069]

[0070] Where Δx wl , Δy wl , Δz wl Represents the physical size adjustment in the horizontal, longitudinal and vertical directions, Δx xs , Δy xs , Δx xs Respectively represent the horizontal, longitudinal and vertical pixel adjustment amounts, W x Indicates the actual width of the horizontal occlusal surface, W y Indicates the actual longitudinal height of the occlusal surface; W z Indicates the actual vertical depth of the occlusal surface, Pit xIndicates the pixel ratio value corresponding to the actual width of the occlusal surface, Pit y Indicates the pixel ratio value corresponding to the actual height of the occlusal surface, Pit z Indicates the pixel ratio value corresponding to the actual depth of the occlusal surface; for example, 1mm corresponds to 10 pixels. If the actual width of the occlusal surface is 15mm, the pixel ratio value is 15×10=150; the width refers to the horizontal dimension of the occlusal surface of the teeth, that is, the distance from one side to the other side of the teeth when they are occluded; the height refers to the vertical distance from the gum line to the occlusal surface of the teeth when they are occluded; the depth refers to the vertical dimension of the occlusal surface of the teeth, that is, the distance from front to back, that is, the size of the contact area between the upper teeth and the lower teeth when the teeth are occluded;

[0071] Mapping the adjustment data converted into physical dimensions to a solid model of the target object to obtain actual adjustment data, wherein the solid model is the actual three-dimensional occlusal plane;

[0072]

[0073] Where Δx act , Δy act , Δz act Respectively represent the actual adjustment amount in the horizontal, longitudinal and vertical directions, and θ represents the inclination angle of the occlusal plane; its purpose is to compensate for the inclination of the occlusal plane and ensure that the direction of the actual adjustment amount is consistent with the anatomical structure and the actual wearing situation;

[0074] The occlusal plane inclination angle refers to the angle between the occlusal plane of the tooth and a reference plane (such as a horizontal plane, a vertical plane or a specific anatomical plane), which reflects the degree of inclination of the tooth in three-dimensional space. It is a key parameter for evaluating the occlusal relationship and can be obtained by importing denture design software (such as exocad) and using a spatial ruler tool to measure the dihedral angle between the occlusal plane and the horizontal plane. It can also be approximately replaced by calculation:

[0075] Among them, H represents the height difference between the posterior teeth and the anterior teeth, L arch Represents the length of the dental arch; it should be noted that the inclination angle of the occlusal plane obtained by this calculation method is not equivalent to the actual inclination angle of the occlusal plane and is only used as an auxiliary parameter.

[0076] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

[0077] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A high-precision digital full denture 3D printing method, characterized in that: The following steps are involved: Step S1: Acquire the oral three-dimensional data of the target object, and obtain the initial denture model of the target object based on the constructed knowledge graph matching; Step S2: Based on the oral three-dimensional data and the corresponding initial denture model, finite element analysis is used to simulate the stress distribution during occlusion, a three-dimensional occlusal heat map is obtained, and the occlusal heat map is input into the constructed CNN network model to obtain the occlusal point adjustment data; Step S3: obtaining the confidence of the occlusal point adjustment data, performing confidence judgment to obtain the effective occlusal contact points, and mapping the adjustment data of the effective occlusal contact points to the physical model to obtain the actual adjustment data; Step S4: The initial denture model is adjusted based on the actual adjustment data, and converted into a 3D printing instruction for 3D printing.

2. A high-precision digital full denture 3D printing method according to claim 1, characterized in that: The oral three-dimensional data in step S1 includes jaw relationship and oral tissue data, wherein the jaw relationship includes vertical jaw relationship and horizontal jaw relationship; the vertical jaw relationship is the vertical distance, and the determination of the horizontal jaw relationship is the determination of the central relationship; the target object is the patient who needs to undergo full denture production.

3. A high-precision digital full denture 3D printing method according to claim 2, characterized in that: The constructed knowledge graph is an oral knowledge graph, including entities, entity attribute information, relationships between entities, and entity types. The entity attribute information is a label describing the entity attributes; the relationship between entities represents the connection between different entities, describing various associations and interactions between entities in the knowledge graph. The entity type defines the category to which the entity belongs, and defines the common attributes and relationships between entities. The oral knowledge graph is constructed in the following manner: The oral 3D data of n edentulous patients are collected in advance, and all of the n edentulous patients need to make full dentures. The 3D models of the finished dentures of the n edentulous patients are obtained, and the oral 3D data of each edentulous patient are formed into a data set, that is, there are n data sets, each data set is an entity in the knowledge graph, and the entity attribute information is the basic information of the edentulous patient corresponding to the entity, and the basic information includes but is not limited to age, gender, name, etc. The entity type is the 3D model of the finished denture of the edentulous patient corresponding to the entity.

4. A high-precision digital full denture 3D printing method according to claim 3, characterized in that: The initial denture model is obtained by: Match the oral three-dimensional data of the target object with the entities in the oral knowledge graph, take the entity with the highest matching degree as the successfully matched entity, mark it as the matching entity, and obtain the entity type corresponding to the matching entity as the initial denture model of the target object; that is, the three-dimensional model of the finished denture of the edentulous patient corresponding to the matching entity is used as the initial denture model of the target object; The method of matching the target object's oral 3D data with the entities in the oral knowledge graph is as follows: The oral three-dimensional data of the target object is formed into a target data set, and the target data set and the data set corresponding to the entities in the oral knowledge graph are preprocessed, and matching is performed by extracting the feature vectors of the data in the data set.

5. A high-precision digital full denture 3D printing method according to claim 4, characterized in that: The three-dimensional occlusal heat map is an RGB image, in which the red channel represents the contact pressure of the anterior teeth area, the green channel represents the contact pressure of the left posterior teeth area, and the blue channel represents the contact pressure of the right posterior teeth area; Obtain occlusal contact points based on the three-dimensional occlusal heat map and mark them as 1, 2, 3, ..., N; input the three-dimensional occlusal heat map into the trained CNN network model, map the N occlusal contact points to the corresponding pixel positions of the output tensor, and output the three-dimensional adjustment amount of each pixel point position; The CNN network model mainly includes an input layer, a convolution layer, a pooling layer, a comprehensive loss function, a fully connected layer and an output layer; the input layer is used to input data, namely, a three-dimensional occlusion heat map; the convolution layer is used to extract primary and advanced features of the data, and the comprehensive loss function combines the position error term, the contact area ratio error term and the smoothing loss; the output layer is used to output data, namely, the three-dimensional adjustment amount of the occlusal contact point and the predicted occlusal contact area ratio.

6. A high-precision digital full denture 3D printing method according to claim 5, characterized in that: The specific method of training the CNN network model is as follows: Based on the pre-collected oral 3D data and 3D occlusal heat map of n edentulous patients, the corresponding denture production data are obtained, and the oral 3D data of each edentulous patient and the corresponding denture production data are used as a group of research data, so there are n groups of research data, which are divided into a training set, a validation set and a test set; Each set of training data is used as the input of the CNN network model, and the predicted bite point adjustment data is used as the output, with the training goal of minimizing the sum of the prediction errors of all training data; the calculation formula of the prediction error is: Where Z k is the prediction error, k is the group number corresponding to the research data, a ki is the occlusal point adjustment data predicted for the i-th occlusal contact point in the k-th group of research data, b ki is the actual occlusal point adjustment data of the i-th occlusal contact point in the k-th group of research data, that is, the occlusal point adjustment data existing in the denture production data; The CNN network model is trained until the sum of the prediction errors reaches convergence, and the training is stopped to obtain a trained CNN network model.

7. A high-precision digital full denture 3D printing method according to claim 6, characterized in that: The formula of the comprehensive loss function is expressed as: Loss total =α·L weizhi +β·L ratio +γ·L smooth ; Among them, Loss total Represents the comprehensive loss function, L weizhi Represents the position error term loss, L ratio Represents the contact area ratio error term loss, L smooth represents smoothing loss; α, β, and γ are the corresponding weight coefficients respectively; in, represents the predicted value of the three-dimensional displacement vector of the i-th occlusal contact point, represents the actual value of the three-dimensional displacement vector of the i-th occlusal contact point; i = 1, 2, 3, ..., N; N is the total number of contact points; in, represents the predicted occlusal contact area ratio, Indicates the actual occlusal area contact ratio.

8. A high-precision digital full denture 3D printing method according to claim 7, characterized in that: The occlusal point adjustment data is the three-dimensional position adjustment amount of the pixel point corresponding to the occlusal contact point, the horizontal adjustment amount is marked as Δx, the longitudinal adjustment amount is marked as Δy, and the vertical adjustment amount is marked as Δz, and the adjustment direction can be represented by a positive or negative sign; The confidence of the occlusal point adjustment data is obtained, that is, the confidence of the pixel corresponding to the occlusal contact point is obtained, and whether the pixel is a valid pixel is determined. If it is a valid pixel, the occlusal contact point corresponding to the pixel is a valid occlusal contact point.

9. A high-precision digital full denture 3D printing method according to claim 8, characterized in that: In step S3, the adjustment data of the effective occlusal contact points are mapped to the physical model, and the specific method of obtaining the actual adjustment data is as follows: Convert the adjustment amount of the effective occlusal contact point into a physical size; The adjustment data converted into physical dimensions is mapped to a solid model of the target object to obtain actual adjustment data, wherein the solid model is the actual three-dimensional occlusal plane.

10. A high-precision digital full denture 3D printing method according to claim 9, characterized in that: The formula for converting the adjustment amount of the effective occlusal contact point into the physical size is expressed as: Where Δx wl , Δy wl , Δz wl Represents the physical size adjustment in the horizontal, longitudinal and vertical directions, Δx xs , Δy xs , Δx xs Respectively represent the horizontal, longitudinal and vertical pixel adjustment amounts, W x Indicates the actual width of the horizontal occlusal surface, W y Indicates the actual longitudinal height of the occlusal surface; W z Indicates the actual vertical depth of the occlusal surface, Pit x Indicates the pixel ratio value corresponding to the actual width of the occlusal surface, Pit y Indicates the pixel ratio value corresponding to the actual height of the occlusal surface, Pit z Indicates the pixel ratio value corresponding to the actual depth of the occlusal surface.

Images