Processing system and processing method of bracket-free invisible orthodontic tooth socket

Abstract

The invention relates to the technical field of orthodontics, and discloses a processing system and a processing method for a bracket-free invisible orthodontic tooth socket, and the processing system comprises a data collection and integration module which is used for collecting and integrating multi-source data, comprising collection approaches and specific operation processes of oral three-dimensional scanning data, basic information of a patient, clinical examination records, correction scheme records and correction result evaluation; the data standardization and cleaning module is used for carrying out format unification and cleaning on the collected data; and the feature extraction and labeling module is used for extracting tooth form and tooth space relation features from the oral cavity three-dimensional scanning data and labeling the malocclusion and occlusion relation in the oral cavity. The invisible orthodontic tooth socket is manufactured by collecting a large amount of multi-dimensional data including oral three-dimensional scanning data, patient clinical information, orthodontic process records, result evaluation and the like, and the moving path of each tooth in each orthodontic stage and the magnitude and direction of required orthodontic force are accurately predicted.

Description

Technical Field

[0001] The present invention relates to the technical field of oral orthodontics, and in particular to a processing system and a processing method for bracketless invisible orthodontic braces. Background Art

[0002] As people pay more and more attention to oral health and aesthetics, the demand for orthodontic treatment is growing. Traditional orthodontic methods, such as metal braces, have problems such as poor aesthetics and low wearing comfort. As an emerging orthodontic treatment method, bracketless invisible orthodontic braces have attracted widespread attention due to their advantages such as beauty, removability, and convenience in oral hygiene maintenance.

[0003] However, the processing of bracketless invisible orthodontic braces currently faces many challenges. In traditional processing schemes, the formulation of correction plans often relies on the experience of orthodontists and simple measuring tools. Doctors design correction plans through visual inspection of the patient's mouth, model analysis and some basic measurements (such as arch length, width, etc.). This method has great limitations, because human experience and judgment cannot fully consider the complex anatomical structure of the patient's mouth, individual differences in teeth and alveolar bones, and various potential influencing factors. For example, simple measurements of the degree of dental crowding may not accurately assess the impact of tooth torsion and inclination on the correction plan, resulting in inaccurate treatment plans, which may lead to problems such as unsatisfactory tooth movement, prolonged treatment time, and even damage to periodontal tissues during the treatment process.

[0004] In terms of braces design, traditional methods usually use universal design templates, which lack sufficient consideration of the patient's individual occlusal relationship, tooth morphology and functional movement. This makes it possible for the braces to not fit perfectly with the tooth surface during wearing, affecting the transmission and distribution of correction force and reducing the correction effect. Moreover, in terms of manufacturing technology, neither hot pressing molding nor some early 3D printing technologies are fully combined with personalized design, and the braces produced have certain defects in quality and correction function.

[0005] In addition, with the continuous increase in orthodontic treatment cases, the amount of data has exploded, but most of this data has not been effectively integrated and utilized. The data formats between different medical institutions are not unified and the quality is uneven, which makes it impossible to form a resource that can guide clinical practice. At the same time, there is a lack of effective data analysis methods, making it difficult to extract valuable information from a large amount of data to improve correction plans and braces processing technology. Summary of the invention

[0006] In view of the deficiencies in the prior art, the present invention provides a processing system and method for bracketless invisible orthodontic braces, which solves the problem of insufficient consideration of the patient's individual occlusal relationship, tooth morphology and functional movement.

[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions: a processing system for bracketless invisible orthodontic braces, comprising:

[0008] The data collection and integration module collects and integrates data from multiple sources, including the collection methods and specific operation processes of oral 3D scanning data, basic patient information, clinical examination records, correction plan records, and correction result evaluation;

[0009] Data standardization and cleaning module, which standardizes and cleans the collected data;

[0010] Feature extraction and annotation module, which extracts tooth morphology and tooth spatial relationship features from oral 3D scanning data, and annotates malocclusion and occlusion relationships in the oral cavity;

[0011] The model building module builds the model through the received oral 3D scanning data, including convolutional neural network layer, graph neural network layer and recurrent neural network layer, which respectively extract different types of features and fuse them to better predict the correction plan;

[0012] Model training module, to train the constructed model;

[0013] The processing module is used to process and produce bracketless invisible orthodontic braces through the constructed model.

[0014] Preferably, the data collection and integration module includes:

[0015] The oral 3D scanning unit uses laser scanning to quickly and non-contactly scan the inside of the cavity, collect point cloud data of the oral surface, and perform preliminary noise and reflection spot processing;

[0016] The patient information collection unit collects information about the patient, including name, age, gender, contact information, race, occupation, family orthodontic history, general health status, and oral hygiene habits;

[0017] Clinical examination record unit, to record the patient's periodontal condition and occlusal relationship;

[0018] The correction plan recording unit specifies the source, size, direction and point of application of the correction force.

[0019] Preferably, the data standardization and cleaning module includes:

[0020] The data format unification unit unifies data from multiple sources into an internal data format, and ensures the consistency of spatial positioning of all data by defining a three-dimensional coordinate system with the anatomical landmarks in the oral cavity as the origin;

[0021] The data cleaning unit uses bilateral filtering algorithm to remove noise;

[0022] The data review unit automatically checks the consistency between the patient's basic information, clinical examination records and correction plan records.

[0023] Preferably, the feature extraction and annotation module includes:

[0024] The geometric feature extraction unit extracts the tooth morphology, tooth spatial relationship, and alveolar bone structure features;

[0025] The feature annotation unit is used to annotate the malocclusion type, tooth occlusion relationship and periodontal health status.

[0026] Preferably, the model building module includes:

[0027] Model architecture design unit, building convolutional neural network layers, graph neural network layers, and recurrent neural network layers;

[0028] Hybrid neural network construction, build a hybrid neural network including convolutional neural network layer, graph neural network layer and recurrent neural network layer.

[0029] Preferably, the model training module includes:

[0030] The training data preparation unit is used to divide the processed data into a training set, a validation set and a test set, wherein the division adopts a stratified sampling method and the training data is enhanced.

[0031] Preferably, the biomechanical research data collected by the data collection and integration module include stress-strain curves of orthodontic appliances made of different materials under different stress conditions, which are obtained by performing tensile, compression and bending mechanical tests on orthodontic appliance samples of standard shapes using a material testing machine.

[0032] Preferably, when the training data preparation unit divides the data set by stratified sampling, the stratification factors include malocclusion type, age, and gender.

[0033] Preferably, the graph neural network layer uses a multi-layer graph convolutional network operation to fuse adjacent node information to learn the mutual influence and relationship between teeth, and the recurrent neural network layer concatenates the features processed by CNN and GNN with the encoding vectors of the patient's basic information and clinical examination records. The encoding vector converts discrete codes into continuous vector representations through an embedding layer.

[0034] In addition, the present invention also provides a method for processing a bracket-free invisible orthodontic brace, comprising the following steps:

[0035] S1. Use a high-precision oral scanner to scan the patient's oral cavity from multiple angles after calibration, covering all surfaces of the upper and lower teeth and gum tissue, and obtain and preliminarily process point cloud data;

[0036] S2. Define the three-dimensional coordinate system of oral data, standardize the point cloud density and color coding, encode the basic information of patients and clinical examination records according to medical standards, manually review and correct fuzzy information, use statistical analysis methods to identify oral scan data noise points and process them with bilateral filtering algorithm;

[0037] S3, using point cloud processing algorithms and geometric operations to extract crown height, width, root length and shape, and cusp angle from oral 3D scanning data, and determine tooth adjacency and measure gap size through point cloud data;

[0038] S4, receiving oral 3D scanning data, extracting local features with a specific convolution kernel, constructing a graph structure with teeth as nodes and adjacency relationships as edges, describing the topological relationship of teeth, concatenating the processed features with the patient information encoding vector, and converting the discrete code into a continuous vector in the embedding layer;

[0039] S5. Select the 3D printing process according to the design requirements and material properties of the orthodontic appliance, set the corresponding materials and parameters, and manufacture bracketless invisible orthodontic braces.

[0040] The present invention provides a processing system and method for bracketless invisible orthodontic braces, which has the following beneficial effects:

[0041] 1. The present invention collects a large amount of multi-dimensional data including oral 3D scanning data, patient clinical information, correction process records and result evaluation, and uses advanced data processing and artificial intelligence models to conduct a comprehensive and detailed analysis of the patient's oral condition. It accurately predicts the movement path of each tooth at each correction stage, the size and direction of the required correction force, and effectively avoids the problem of poor correction effect or complications caused by inaccurate solutions.

[0042] 2. The present invention integrates rich patient data and analyzes it based on a complex neural network architecture, taking into full account individual differences in patients when generating treatment plans. Factors such as age, gender, race, general health status, oral hygiene habits, type and severity of malocclusion of different patients are taken into consideration, so that each patient can receive the orthodontic treatment that best suits their situation, improving the patient's comfort and compliance with treatment, and also helping to improve the treatment effect.

[0043] 3. The present invention standardizes and cleans a large amount of data from multiple oral medical institutions, removes erroneous, incomplete and inconsistent data, and improves data quality. Through feature extraction and annotation, valuable information is mined from the original data, which not only provides a basis for the formulation of the current patient's correction plan, but also provides rich resources for subsequent research and improvement. Moreover, using artificial intelligence models to learn and train these data can discover the complex patterns and laws hidden in the data, and further optimize the generation of correction plans.

[0044] 4. The present invention has clear methods and specifications for each link, from data collection, analysis, correction plan generation to braces design and manufacturing, as well as dynamic adjustment during the correction process. This helps different oral medical institutions and orthodontists follow unified standards during the treatment process, reducing the problem of uneven treatment levels caused by human factors or differences between different institutions. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 It is the architecture diagram of the processing system in the present invention. DETAILED DESCRIPTION

[0046] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0047] Please see attached Figure 1 The embodiment of the present invention provides a processing system and a processing method for bracketless invisible orthodontic braces, comprising:

[0048] 1. Data collection and integration module

[0049] In this embodiment, the data collection and integration module collects and integrates data from multiple sources, including oral 3D scanning data, basic patient information, clinical examination records, correction plan records, and correction result evaluation. The collection methods and specific operation processes are as follows:

[0050] 1.1. Oral 3D Scanning Unit

[0051] In this embodiment, when a high-precision oral scanner is used to collect oral three-dimensional model data, the scanner is first calibrated to ensure scanning precision and accuracy. The calibration process includes checking and adjusting the optical system and measurement accuracy of the scanner, scanning using a standard calibration model, comparing the scanning results with the actual parameters of the standard model, and adjusting the focal length and light intensity parameters of the scanner until the scanning error is within the allowable range (less than 0.1mm). During the scanning process, the patient's mouth is scanned from multiple angles, and the scanning angle covers the entire mouth, including the labial and buccal surfaces, lingual and palatal surfaces, occlusal surfaces, and gingival tissue of the upper and lower teeth. The scanning head of the scanner moves around the patient's mouth and scans according to the preset scanning path and angle. The settings of the scanning path and angle are optimized to ensure that detailed data of all parts of the mouth are obtained without omission. During the scanning process, the scanner acquires point cloud data of the oral surface in real time, and performs preliminary processing on the data, such as removing obvious noise and reflected light spots;

[0052] 1.2. Patient information collection unit

[0053] In this embodiment, the patient's basic information covers a wide range of content, such as name, age, gender, contact information, race, occupation, family orthodontic history, general health status: whether the patient suffers from diabetes, heart disease, immune system disease and other diseases that may affect orthodontic treatment, and detailed inquiries and records of disease type, severity and treatment, oral hygiene habits: brushing frequency, use of dental floss, whether to use mouthwash). This information is recorded in detail by medical staff in electronic forms when the patient is first diagnosed;

[0054] 1.3. Clinical examination record unit

[0055] In this embodiment, the periodontal condition examination uses a periodontal probe to measure indicators such as the gingival sulcus depth and gingival bleeding index to evaluate the gingival health. The occlusal relationship examination uses articulating paper to determine the occlusal contact points, measure the overbite and overjet, and classify the occlusion type according to the Angle classification method. At the same time, the doctor will also check the function of the temporomandibular joint and determine the joint health by palpation, observation of the opening degree and type of opening. For tooth morphology and arrangement, record the number of teeth, morphological abnormalities, and the degree of crowding or sparseness of the teeth. All these examination results are entered into the electronic medical record system in standardized text and digital form;

[0056] 1.4. Correction plan recording unit

[0057] Formulated by orthodontists based on the patient's diagnosis. The treatment goals are clear, such as improving occlusal function, aligning teeth, and adjusting facial aesthetics. The treatment steps are planned in detail, including dividing the entire treatment process into several stages, each stage has specific treatment tasks, such as moving certain teeth, adjusting occlusal relationships, etc. The type of appliance is selected as bracketless invisible orthodontic braces, and the reasons for the selection are explained. The expected course of treatment is estimated in months, and the design of the correction force system is described in detail, that is, for each step of correction, the source of the correction force (such as elastic deformation of the braces, assistance of accessories), size, direction and point of action (specific location on the tooth surface) are clearly defined. This information is recorded in a special correction plan document in the form of text and charts, and the evaluation of the correction results is carried out after the correction is completed. Oral 3D scanning, clinical examination and imaging examination (such as curved surface tomography, lateral head film) are performed again. The correction effect is evaluated by comparing the data before and after correction. The improvement of tooth arrangement is evaluated using the PAR index method, and the change in the degree of tooth disorder before and after correction is calculated. The restoration of occlusal function is determined by checking the uniformity and stability of the occlusal contact and whether the ideal occlusal relationship is achieved. The facial aesthetic changes are evaluated by comparing the front and side photos to subjectively and objectively evaluate the improvement of facial contour and proportion. The health status of periodontal tissues is checked to see whether the gums are inflamed or retracted, and whether the alveolar bone height is stable. These evaluation results are recorded in the form of detailed reports to provide a basis for subsequent data annotation and model training;

[0058] 2. Data standardization and cleaning module

[0059] In this embodiment, the data standardization and cleaning module standardizes and cleans the collected data, as follows:

[0060] 2.1. Data format unified unit

[0061] In this embodiment, a three-dimensional coordinate system is defined with a specific anatomical landmark point in the oral cavity as the origin (such as a point on the midsagittal plane) to ensure the consistency of spatial positioning of all data. At the same time, the data resolution is standardized and the point cloud density of each tooth surface is specified for subsequent processing. If the scanned data contains color information (used to distinguish different tissue structures), the color coding is also standardized. During the conversion process, the data is resampled to keep the data resolution of all cases consistent, which is convenient for subsequent feature extraction and model processing. The text information in the patient's basic information and clinical examination records is encoded according to medical terminology standards. For example, the disease names in the patient's general health status are converted using an internationally accepted medical classification system. For the brushing frequency options in oral hygiene habits (such as once a day, twice a day), corresponding digital codes are also assigned. For some vague or non-standard expressions, professional medical information personnel will manually review and correct them to ensure the accuracy and consistency of the data;

[0062] 2.2 Data Cleaning Unit

[0063] In this embodiment, noise removal in oral scan data is a key step. First, a statistical analysis method is used to identify abnormal points. By calculating the statistical characteristics of the distance between each point and its neighboring points (such as average distance, standard deviation), the points whose distance exceeds a certain threshold are determined to be noise points. For these noise points, a bilateral filtering algorithm is used for processing. The bilateral filtering algorithm takes into account both spatial distance and data value differences, and can retain the edge and detail information of the data while removing noise points. For artifacts of larger areas (such as false images caused by reflections or foreign objects in the oral cavity during scanning), a region growing algorithm is used to identify and remove them. The region growing algorithm starts with a seed point (a point in the normal data area selected manually or automatically) and gradually expands according to similarity criteria (such as similar geometric features such as the normal vector direction and curvature of the point) until the entire artifact area is covered, and then it is deleted to ensure the quality of the scan data;

[0064] 2.3 Data Review Unit

[0065] Develop a data review algorithm to automatically check the consistency between the patient's basic information, clinical examination records, and treatment plan records. For example, if the patient's age is shown as a teenager, but serious geriatric diseases are recorded in the overall health status, or the correction goals set in the treatment plan are obviously inconsistent with the type of malocclusion in the clinical examination, the algorithm will mark these abnormalities. For the marked contradictory information, a team of professional orthodontists and medical information personnel will conduct manual review. According to the patient's actual situation and medical logic, the erroneous information will be corrected or deleted. If there are any questions, further communication and verification will be carried out with the medical staff who collected the data to ensure the integrity and reliability of the data;

[0066] 3. Feature extraction and annotation module

[0067] In this embodiment, tooth morphology and tooth spatial relationship features are extracted from the oral 3D scanning data, and the malocclusion and occlusion relationship in the oral cavity are annotated, as follows:

[0068] 3.1 Geometric Feature Extraction Unit

[0069] Extraction of tooth morphological features: Extracting the morphological features of teeth from oral 3D scanning data is a complex but critical process. For crown height, the distance between the cusp vertex and the lowest point of the gingival margin is accurately found in the direction of the long axis of the tooth, and the distance between the two points is calculated. This requires a deep understanding of the geometric structure of the tooth and the use of point cloud processing algorithms in three-dimensional space. The crown width is achieved by measuring the distance at the widest point of the crown (on a plane perpendicular to the long axis of the tooth) and analyzing the point cloud data of the tooth surface on a specific plane. The extraction of root length and morphology is achieved by curve fitting of the point cloud data of the root part, such as fitting a quadratic curve or a cubic curve using the least squares method, and then calculating the curve length and curvature changes to describe the characteristics of the root in detail. The cusp angle is determined by calculating the angle between the normal vectors of the adjacent surfaces at the cusp vertex, which involves vector operations and surface analysis in three-dimensional geometry. These calculations are based on a mature three-dimensional point cloud processing algorithm library to ensure the accuracy of feature extraction;

[0070] Calculation of spatial relationship characteristics of teeth: The spatial relationship characteristics between teeth are crucial for the formulation of orthodontic treatment plans. When calculating the adjacent relationship between teeth, the adjacency is determined by judging whether the minimum distance between the surface point clouds of two teeth is less than a certain threshold (determined according to the tooth anatomy standards). The size of the tooth gap is determined by measuring the shortest distance between adjacent teeth (in the direction perpendicular to the adjacent surface), and the point cloud data of adjacent teeth needs to be accurately analyzed. For the distribution of occlusal contact points, the occlusion of the three-dimensional models of the upper and lower teeth is simulated. By moving and rotating the mandibular model within a certain range, the best occlusal position is found (the iterative closest point algorithm can be used for model alignment), and then the points where the upper and lower tooth surfaces contact are marked, and the number, position and distribution of the contact points are counted to provide a basis for analyzing the occlusal function and formulating correction plans;

[0071] Assessment of alveolar bone structural characteristics: The structural characteristics of the alveolar bone have an important impact on the safety and effectiveness of orthodontic treatment. For the height of the alveolar bone, it is measured by the distance from the alveolar ridge top to the root tip in the direction of the long axis of the tooth (combined with the segmentation algorithm of the tooth and alveolar bone, the point clouds of the two are processed separately to improve the accuracy of the measurement). The thickness of the alveolar bone is evaluated by measuring the shortest distance from the alveolar bone surface to the root surface in different directions around the root, which requires a complex three-dimensional spatial measurement algorithm. Alveolar bone density information can be obtained by mapping the density information in the imaging to the three-dimensional scanning data through registration with CBCT (cone beam CT) imaging data (using image registration algorithms, such as mutual information-based registration methods), or through indirect methods (such as estimation based on the color, texture and other characteristics of the alveolar bone) to fully understand the condition of the alveolar bone;

[0072] 3.2 Feature Annotation Unit

[0073] Malocclusion type annotation: More detailed annotation is performed based on the Angle classification method combined with the results of three-dimensional occlusal analysis. For Angle Class I malocclusion, further annotation is performed to determine whether there is crowding or sparse dentition, as well as the degree of crowding or sparse dentition. A new crowding evaluation formula is used:

[0074] C Inew =CI×(1+μ)

[0075] Among them, CI is the traditional crowding index, which is obtained by calculating the difference ratio between the expected length of the dental arch and the current length, and μ is calculated by trigonometric function according to the tooth rotation angle and inclination degree. The rotation and inclination calculation of the tooth is determined by analyzing the angle between the long axis of the tooth and the ideal dental arch plane and the change of the normal vector of the crown surface. For Angle Class II and Class III malocclusion, it is marked whether it is dental or skeletal malocclusion (by analyzing the morphology and positional relationship of the jaw, combined with imaging data such as lateral skull radiographs, if available), and the severity of the malocclusion (based on the degree of difference between the upper and lower jaws in the anterior-posterior and vertical directions, using the ANB angle measurement index), to provide a detailed basis for the formulation of the correction plan;

[0076] Occlusal relationship annotation: In addition to recording the values ​​of overbite and overjet (by measuring the vertical and horizontal distances between the incisor edges of the upper and lower anterior teeth), the stability and functionality of the occlusion are also annotated. By observing the changes in the occlusal contact points in different functional movements (such as chewing, protrusion, and lateral movement), it is possible to determine whether the occlusion is stable and whether there are problems such as early contact and interference. This information is annotated in the data in a coded form, for example, using numbers to represent different occlusal types (0-normal, 1-deep overbite, 2-open bite, etc.), and using letters to represent the stability of the occlusion (A-stable, B-unstable) and functionality (F-good function, P-functional problems), so that the model can learn the complex characteristics of the occlusal relationship;

[0077] Periodontal health status annotation: According to the periodontal examination results, the gum health status is annotated into different levels (such as 0-healthy, 1-mild inflammation, 2-moderate inflammation, 3-severe inflammation, judged by gum color, texture, bleeding and other indicators). The alveolar bone health status is annotated as stable, mild absorption, moderate absorption, and severe absorption (based on the measurement results of alveolar bone height and a comprehensive analysis of factors such as age and general health status). These annotation information is associated with the patient's basic information and clinical examination records, so that the condition of the periodontal tissue can be fully considered in model training and analysis, and a more appropriate correction plan can be formulated;

[0078] 4. Model building module

[0079] In this embodiment, a model is constructed by receiving the oral 3D scanning data, including a convolutional neural network layer, a graph neural network layer, and a recurrent neural network layer, respectively extracting different types of features and fusing them to better predict the correction plan, as follows:

[0080] Convolutional Neural Network (CNN) Layer: The input layer of the model receives the standardized three-dimensional oral scan data. The CNN layer first uses a series of convolutional layers to extract local features. The convolution kernel size is selected as 3*3 (considering three-dimensional data), the step size is 1, and the padding method is 'SAME' (to ensure that the output feature map size is the same as the input). This setting enables the convolution kernel to fully cover the local area of ​​the data and extract rich feature information when sliding on the scanned data. For example, different convolution kernels can learn different patterns for features such as texture and edges on the tooth surface. Each convolution kernel extracts specific local features by learning different weights. For example, some convolution kernels may focus on extracting tiny bumps on the tooth surface, while others are better at capturing changes in the sharpness of the tooth edges;

[0081] Graph Neural Network (GNN) layer: The teeth in the oral 3D scan data are regarded as nodes in the graph, and the adjacency relationship between teeth is regarded as edges, to build a graph structure that reflects the oral tooth structure. Specifically, an adjacency matrix A is constructed. ij , where A ij Indicates whether tooth i and tooth j are adjacent (1 for adjacent, 0 for not adjacent). At the same time, the feature vector of each tooth extracted from the CNN layer is used as the initial feature representation X of the node. It can accurately describe the topological relationship between teeth;

[0082] Recurrent Neural Network (RNN) layer: concatenates the features processed by CNN and GNN with the encoding vectors of the patient's basic information and clinical examination records. These encoding vectors are converted from discrete encodings to continuous vector representations through the embedding layer. The embedding layer learns a low-dimensional vector representation for each discrete encoding value, allowing the model to better process this information. For example, for information such as the patient's different genders and age ranges, the embedding layer will convert it into a vector with certain semantic information, which is convenient for network learning and processing;

[0083] 5. Model training module

[0084] In this embodiment, the constructed model is trained as follows:

[0085] 5.1. Training data preparation unit

[0086] In this embodiment, after completing data collection, integration, standardization and feature extraction, the data is divided into a training set, a validation set and a test set. In this embodiment, the division is performed in a ratio of approximately 70%, 20%, and 10%. In the division process, ensure that each subset is representative in terms of various malocclusion types, age distribution, gender ratio, etc. The labeled information of the divided data set is checked again to ensure that the labeling of each data sample is accurate, especially for key contents such as malocclusion type, tooth movement information, correction force related labeling, etc., which are checked by combining manual sampling and automatic verification algorithms;

[0087] Enhanced processing: Perform small random perturbations on clinical parameters (such as crowding and occlusal relationship). Taking crowding as an example, based on its original value, a perturbed crowding value is generated at a ratio of ±10%. This perturbation simulates the error in actual clinical measurements and improves the robustness of the model to parameter changes;

[0088] 6. Processing module

[0089] In this embodiment, the invisible orthodontic sleeve without brackets is processed by the constructed model, as follows: according to the design requirements and material properties of the appliance, a suitable processing technology is selected, such as hot pressing or 3D printing. For the hot pressing process, a suitable hot pressing material (such as a specific type of medical polymer) is selected, and the hot pressing temperature, pressure and time parameters are set according to the thermal performance parameters of the material (such as melting point, glass transition temperature). During the hot pressing process, the preformed material sheet is placed in the mold, and the shape of the mold is made according to the designed braces shape. By precisely controlling the hot pressing equipment, the material fits the mold shape under high temperature and high pressure to form the braces. For the 3D printing process, a suitable printing material (such as a light-curing resin with good biocompatibility and mechanical properties) is selected, and the printing parameters are set, including layer thickness (such as 0.1mm), printing speed (such as 50mm / s), and filling density (such as 80%). The 3D printing equipment is used to stack the materials layer by layer according to the designed braces model data to manufacture the appliance.

[0090] In addition, an embodiment of the present invention also provides a method for processing a bracketless invisible orthodontic brace, comprising the following steps:

[0091] S1. Use a high-precision oral scanner, calibrate it and scan the patient's oral cavity from multiple angles, cover all surfaces of the upper and lower teeth and gum tissue, obtain and preliminarily process point cloud data, and medical staff use electronic spreadsheets to record basic information such as name and age, including general health status and oral hygiene habits during the initial visit of the patient, and formulate a plan including correction goals, steps, type of orthodontic appliance, expected course of treatment and correction force system design. After the correction is completed, various examinations are performed to evaluate the effect and record it;

[0092] S2. Define the three-dimensional coordinate system of oral data, standardize the point cloud density and color coding, encode the basic information of patients and clinical examination records according to medical standards, manually review and correct fuzzy information, use statistical analysis methods to identify oral scan data noise points and process them with bilateral filtering algorithms, use regional growing algorithms to remove large-area artifacts, develop data review algorithms to automatically check data consistency, and have professionals manually review and communicate and verify the contradictory information;

[0093] S3. Extract crown height, width, root length and shape, and cusp angle from oral 3D scanning data using point cloud processing algorithms and geometric operations. Determine tooth adjacency and measure gap size using point cloud data. Simulate occlusion through model registration to find contact point distribution. Measure alveolar bone height and thickness using segmentation algorithms. Evaluate alveolar bone density using CBCT registration. Annotate malocclusion type, degree of crowding or sparseness (for Angle Class I), dental or skeletal nature, and severity using Angle classification combined with 3D occlusal analysis.

[0094] S4, receiving oral 3D scanning data, extracting local features with a specific convolution kernel, constructing a graph structure with teeth as nodes and adjacency relationships as edges, describing the topological relationship of teeth, concatenating the processed features with the patient information encoding vector, and converting the discrete code into a continuous vector in the embedding layer;

[0095] S5. Select the 3D printing process according to the design requirements and material properties of the orthodontic appliance, set the corresponding materials and parameters, and manufacture bracketless invisible orthodontic braces.

[0096] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A processing system for bracketless invisible orthodontic braces, characterized in that: include: The data collection and integration module collects and integrates data from multiple sources, including the collection methods and specific operation processes of oral 3D scanning data, basic patient information, clinical examination records, correction plan records, and correction result evaluation; Data standardization and cleaning module, which standardizes and cleans the collected data; Feature extraction and annotation module, which extracts tooth morphology and tooth spatial relationship features from oral 3D scanning data, and annotates malocclusion and occlusion relationships in the oral cavity; The model building module builds the model through the received oral 3D scanning data, including convolutional neural network layer, graph neural network layer and recurrent neural network layer, which respectively extract different types of features and fuse them to better predict the correction plan; Model training module, to train the constructed model; The processing module is used to process and produce bracketless invisible orthodontic braces through the constructed model.

2. A processing system for bracketless invisible orthodontic braces according to claim 1, characterized in that: The data collection and integration module includes: The oral 3D scanning unit uses laser scanning to quickly and non-contactly scan the inside of the cavity, collect point cloud data of the oral surface, and perform preliminary noise and reflection spot processing; The patient information collection unit collects information about the patient, including name, age, gender, contact information, race, occupation, family orthodontic history, general health status, and oral hygiene habits; Clinical examination record unit, to record the patient's periodontal condition and occlusal relationship; The correction plan recording unit specifies the source, size, direction and point of application of the correction force.

3. A processing system for bracketless invisible orthodontic braces according to claim 1, characterized in that: The data standardization and cleaning module includes: The data format unification unit unifies data from multiple sources into an internal data format, and ensures the consistency of spatial positioning of all data by defining a three-dimensional coordinate system with the anatomical landmarks in the oral cavity as the origin; The data cleaning unit uses bilateral filtering algorithm to remove noise; The data review unit automatically checks the consistency between the patient's basic information, clinical examination records and correction plan records.

4. A processing system for bracketless invisible orthodontic braces according to claim 1, characterized in that: The feature extraction and annotation module includes: The geometric feature extraction unit extracts the tooth morphology, tooth spatial relationship, and alveolar bone structure features; The feature annotation unit is used to annotate the malocclusion type, tooth occlusion relationship and periodontal health status.

5. The processing system of a bracketless invisible orthodontic braces according to claim 1, characterized in that: The model building module includes: Model architecture design unit, building convolutional neural network layers, graph neural network layers, and recurrent neural network layers; Hybrid neural network construction, build a hybrid neural network including convolutional neural network layer, graph neural network layer and recurrent neural network layer.

6. A processing system for bracketless invisible orthodontic braces according to claim 1, characterized in that: The model training module includes: The training data preparation unit is used to divide the processed data into a training set, a validation set and a test set, wherein the division adopts a stratified sampling method and the training data is enhanced.

7. A processing system for bracketless invisible orthodontic braces according to claim 1, characterized in that: The biomechanical research data collected by the data collection and integration module include stress-strain curves of orthodontic appliances made of different materials under different stress conditions, which are obtained by performing tensile, compression and bending mechanical tests on orthodontic appliance samples of standard shapes using a material testing machine.

8. The processing system of a bracketless invisible orthodontic braces according to claim 6, characterized in that: When the training data preparation unit divides the data set by stratified sampling, the stratification factors include malocclusion type, age, and gender.

9. A processing system for bracketless invisible orthodontic braces according to claim 5, characterized in that: The graph neural network layer uses a multi-layer graph convolutional network operation to fuse adjacent node information to learn the mutual influence and relationship between teeth. The recurrent neural network layer concatenates the features processed by CNN and GNN with the encoding vectors of the patient's basic information and clinical examination records. The encoding vector converts discrete codes into continuous vector representations through an embedding layer.

10. A method for processing bracketless invisible orthodontic braces, applied to a processing system for bracketless invisible orthodontic braces according to any one of claims 1 to 9, characterized in that: The following steps are involved: S1. Use a high-precision oral scanner to scan the patient's oral cavity from multiple angles after calibration, covering all surfaces of the upper and lower teeth and gum tissue, and obtain and preliminarily process point cloud data; S2. Define the three-dimensional coordinate system of oral data, standardize the point cloud density and color coding, encode the basic information of patients and clinical examination records according to medical standards, manually review and correct fuzzy information, use statistical analysis methods to identify oral scan data noise points and process them with bilateral filtering algorithm; S3, using point cloud processing algorithms and geometric operations to extract crown height, width, root length and shape, and cusp angle from oral 3D scanning data, and determine tooth adjacency and measure gap size through point cloud data; S4, receiving oral 3D scanning data, extracting local features with a specific convolution kernel, constructing a graph structure with teeth as nodes and adjacency relationships as edges, describing the topological relationship of teeth, concatenating the processed features with the patient information encoding vector, and converting the discrete code into a continuous vector in the embedding layer; S5. Select the 3D printing process according to the design requirements and material properties of the orthodontic appliance, set the corresponding materials and parameters, and manufacture bracketless invisible orthodontic braces.

Images