Orthodontic image classification model determination system based on deep learning

Abstract

The invention discloses an orthodontic image classification model determination system based on deep learning, and relates to the technical field of image data, a construction module is used for constructing an orthodontic data annotation library, the orthodontic data annotation library comprises a plurality of tooth superposition models and a plurality of upper and lower jaw superposition models, and an orthodontic sample image is annotated according to the orthodontic data annotation library; and a training module connected with the construction module and used for determining feature information of the marked orthodontic sample image, selecting a deep learning model according to the feature information, and training the selected deep learning model through the marked orthodontic sample image to obtain a trained deep learning model. According to the method, data integration and labeling can be carried out on many tooth problems through the orthodontic data labeling library, the accuracy and consistency of large-scale data coverage and image labeling can be realized, oral cavity related images of personnel can be quickly labeled, and the orthodontic analysis efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the field of image data technology, and in particular to an orthodontic image classification model determination system based on deep learning. Background Art

[0002] With the development of social economy and the improvement of people's living standards, the public's attention to oral health and aesthetics is increasing. More and more people realize that orthodontic treatment can not only improve problems such as uneven teeth, but also have a positive impact on chewing function, pronunciation and facial aesthetics. Therefore, the demand for orthodontic treatment is constantly increasing. In the preparation process for orthodontic treatment, it is necessary to analyze the teeth and the upper and lower jaw conditions of the person. The analysis is based on oral X-rays and pictures of the inside of the mouth. Deep learning, as an important branch of artificial intelligence, has achieved remarkable results in image recognition, classification and other fields. At present, deep learning has also participated in orthodontic auxiliary diagnosis. On this basis, accurate image annotation is the basis of model training. However, the annotation of orthodontic images requires professional orthodontists or experts in related fields. The annotation process is time-consuming and labor-intensive, and there may be problems with inconsistent annotations. Summary of the Invention

[0003] The purpose of the present invention is to provide an orthodontic image classification model determination system based on deep learning to address the shortcomings of the background technology.

[0004] In order to achieve the above objectives, the present invention provides the following technical solution: an orthodontic image classification model determination system based on deep learning, comprising:

[0005] A construction module is used to construct an orthodontic data annotation library, wherein the orthodontic data annotation library includes multiple tooth superposition models and multiple upper and lower jaw superposition models, and annotate orthodontic sample images according to the orthodontic data annotation library;

[0006] The training module is connected to the construction module and is used to determine the feature information after annotation, select a deep learning model based on the feature information, and train the selected deep learning model using the annotated orthodontic sample images to obtain a trained deep learning model;

[0007] The evaluation module is connected to the training module and is used to evaluate the trained deep learning model through the verification data set. When the accuracy of the trained deep learning model meets the preset conditions, it is used as the target model.

[0008] In a preferred embodiment, the building blocks include:

[0009] A construction unit is used to construct a standard orthodontic model, wherein the standard orthodontic model includes upper and lower jaw standard models;

[0010] The association unit is used to independently establish the association relationship between the teeth in the maxillary dentition and the mandibular dentition for the single tooth models in the standard orthodontic model;

[0011] A configuration unit is used to configure a space ball for each standard tooth model corresponding to a single tooth in a standard orthodontic model, and to superimpose the single tooth according to the space ball to obtain multiple tooth superimposed models;

[0012] A position changing unit is used to superimpose the upper and lower jaws in a standard orthodontic model to obtain multiple upper and lower jaw superimposed models;

[0013] A binding unit is used to bind multiple tooth superposition models and multiple upper and lower jaw superposition models to corresponding standard orthodontic models respectively, and to perform image annotation on the multiple tooth superposition models and the multiple upper and lower jaw superposition models respectively to obtain an orthodontic data annotation library;

[0014] The annotation unit is used to annotate orthodontic sample images according to the orthodontic data annotation library.

[0015] In a preferred embodiment, the building block comprises:

[0016] The first model building unit is used to collect appearance information of the standard upper and lower jaws and build three-dimensional models of the upper and lower jaws based on the appearance information of the upper and lower jaws;

[0017] The second model building unit is used to collect appearance information of standard teeth and build a three-dimensional model of each single tooth according to the appearance information of the standard teeth;

[0018] A combined construction unit is used to combine the three-dimensional models of all individual teeth with the corresponding three-dimensional models of the upper and lower jaws to construct an orthodontic model;

[0019] The contour binding unit is used to cover multiple adjustment points on the external contour surface of the orthodontic model, bind the adjustment points to the external contour surface of the orthodontic model, and obtain a standard orthodontic model.

[0020] In a preferred embodiment, the association unit includes:

[0021] Independent unit, used to separate the 3D models of all individual teeth in the standard orthodontic model from the 3D models of the upper and lower jaws;

[0022] The arrangement association unit is used to sequentially associate the three-dimensional models of single teeth in the maxillary dentition and the mandibular dentition according to the arrangement order of the standard teeth.

[0023] In a preferred embodiment, the configuration unit includes:

[0024] The copying unit is used to keep the standard orthodontic model unchanged and copy a standard orthodontic model as an adjustment model, and the adjustment model and the standard orthodontic model are in a superimposed state;

[0025] A space configuration unit is configured to configure a space sphere corresponding to each standard tooth model in the adjustment model, and perform angle transformation on the standard tooth model of a single tooth in the space sphere based on a root point according to a preset orientation to obtain multiple transformed standard tooth models;

[0026] The transformation unit is used to configure the corresponding model space of multiple transformed standard tooth models respectively, connect the model spaces of multiple transformed standard tooth models to each other, and superimpose the multiple transformed standard tooth models through the model space according to the root point to obtain multiple tooth superimposed models.

[0027] In a preferred embodiment, the space configuration unit includes:

[0028] a delineation unit, configured to use the lowest point of the tooth root of each standard tooth model in the adjustment model as a root point, and to delineate a spatial sphere of a preset diameter using the root point as a center point for each standard tooth model;

[0029] The transformation unit is used to define a preset orientation of the standard tooth model corresponding to each tooth in the space sphere, and use the root point as the rotation point to transform the standard tooth model within the preset orientation according to a preset difference angle to obtain multiple transformed standard tooth models.

[0030] In a preferred embodiment, the marking unit includes:

[0031] A sample construction unit, used for performing three-dimensional construction of an orthodontic sample image to obtain a sample model;

[0032] An attachment adjustment unit, configured to attach the contour of the sample model via adjustment points on the adjustment model, adjust the contour of the adjustment model, and overlap the adjustment model with the sample model;

[0033] A matching unit, configured to obtain the transformed upper and lower jaw models and the transformed standard tooth model in the adjustment model with the highest degree of overlap corresponding to the sample model;

[0034] The image annotation unit is used to use the image annotations of the transformed upper and lower jaw models and the transformed standard tooth model as the image annotations of the sample model.

[0035] In a preferred embodiment, the training module includes:

[0036] An information acquisition unit, configured to determine characteristic information of the orthodontic sample image after annotation, wherein the characteristic information is the data size of the orthodontic sample image;

[0037] A selection unit, used to select a deep learning model based on feature information;

[0038] The training unit is used to divide the annotated orthodontic sample images into a training data set and a validation data set, and train the selected deep learning model through the training data set to obtain a trained deep learning model.

[0039] In the above technical solution, the technical effects and advantages provided by the present invention are:

[0040] The present invention can realize data integration and annotation of various dental problems through the orthodontic data annotation library, can achieve large-scale data coverage and accuracy and consistency of image annotation, can quickly annotate people's oral-related images, and improve the efficiency of orthodontic analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments described in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0042] Figure 1 This is a system block diagram of the present invention. DETAILED DESCRIPTION

[0043] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of 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. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0044] Example 1, please refer to Figure 1 As shown, the orthodontic image classification model determination system based on deep learning described in this embodiment includes:

[0045] A construction module is used to construct an orthodontic data annotation library, wherein the orthodontic data annotation library includes multiple tooth superposition models and multiple upper and lower jaw superposition models, and annotate orthodontic sample images according to the orthodontic data annotation library;

[0046] The training module is connected to the construction module and is used to determine the feature information after annotation, select a deep learning model based on the feature information, and train the selected deep learning model using the annotated orthodontic sample images to obtain a trained deep learning model;

[0047] The evaluation module is connected to the training module and is used to evaluate the trained deep learning model through the verification data set. When the accuracy of the trained deep learning model meets the preset conditions, it is used as the target model.

[0048] In one embodiment, the building blocks include:

[0049] A construction unit is used to construct a standard orthodontic model, wherein the standard orthodontic model includes upper and lower jaw standard models;

[0050] The association unit is used to independently establish the association relationship between the teeth in the maxillary dentition and the mandibular dentition for the single tooth models in the standard orthodontic model;

[0051] A configuration unit is used to configure a space ball for each standard tooth model corresponding to a single tooth in a standard orthodontic model, and to superimpose the single tooth according to the space ball to obtain multiple tooth superimposed models;

[0052] A position changing unit is used to superimpose the upper and lower jaws in a standard orthodontic model to obtain multiple upper and lower jaw superimposed models;

[0053] A binding unit is used to bind multiple tooth superposition models and multiple upper and lower jaw superposition models to corresponding standard orthodontic models respectively, and to perform image annotation on the multiple tooth superposition models and the multiple upper and lower jaw superposition models respectively to obtain an orthodontic data annotation library;

[0054] The annotation unit is used to annotate orthodontic sample images according to the orthodontic data annotation library.

[0055] In one embodiment, the building block comprises:

[0056] The first model building unit is used to collect appearance information of the standard upper and lower jaws and build three-dimensional models of the upper and lower jaws based on the appearance information of the upper and lower jaws;

[0057] The second model building unit is used to collect appearance information of standard teeth and build a three-dimensional model of each single tooth according to the appearance information of the standard teeth;

[0058] A combined construction unit is used to combine the three-dimensional models of all individual teeth with the corresponding three-dimensional models of the upper and lower jaws to construct an orthodontic model;

[0059] A contour binding unit is used to cover a plurality of adjustment points on the outer contour surface of the orthodontic model, and bind the adjustment points to the outer contour surface of the orthodontic model to obtain a standard orthodontic model;

[0060] In one embodiment, the association unit includes:

[0061] Independent unit, used to separate the 3D models of all individual teeth in the standard orthodontic model from the 3D models of the upper and lower jaws;

[0062] an arrangement association unit, for sequentially associating the three-dimensional models of individual teeth in the maxillary dentition and the mandibular dentition according to the arrangement order of the standard teeth;

[0063] In one embodiment, the configuration unit includes:

[0064] The copying unit is used to keep the standard orthodontic model unchanged and copy a standard orthodontic model as an adjustment model, and the adjustment model and the standard orthodontic model are in a superimposed state;

[0065] A space configuration unit is configured to configure a space sphere corresponding to each standard tooth model in the adjustment model, and perform angle transformation on the standard tooth model of a single tooth in the space sphere based on a root point according to a preset orientation to obtain multiple transformed standard tooth models;

[0066] a transformation unit configured to respectively configure the multiple transformed standard tooth models into corresponding model spaces, connect the model spaces of the multiple transformed standard tooth models to each other, and superimpose the multiple transformed standard tooth models through the model spaces according to a root point to obtain multiple superimposed tooth models;

[0067] In one embodiment, the space configuration unit includes:

[0068] a delineation unit, configured to use the lowest point of the tooth root of each standard tooth model in the adjustment model as a root point, and to delineate a spatial sphere of a preset diameter using the root point as a center point for each standard tooth model;

[0069] The transformation unit is used to define a preset orientation of the standard tooth model corresponding to each tooth in the space sphere, and use the root point as the rotation point to transform the standard tooth model within the preset orientation according to a preset difference angle to obtain multiple transformed standard tooth models.

[0070] It should be noted that orthodontic treatment not only involves the arrangement and movement of teeth, but also needs to consider factors such as the morphology and positional relationship of the maxillary and mandibular bones. For example, some orthodontic problems may be caused by the uncoordinated development of the maxillary and mandibular bones, such as underbite (mandibular protrusion) and underbite (maxillary protrusion). For these cases, observing and analyzing the images of the maxillary and mandibular bones is crucial for accurate diagnosis and formulating treatment plans. Therefore, orthodontic images will include X-rays and CT scan images of the upper and lower jaws to comprehensively evaluate orthodontic problems. At the same time, detailed images of teeth are also an important component, which are used to observe the morphology, position, inclination, occlusal relationship, etc. of the teeth, such as periapical films of teeth, panoramic oral films, etc. It can provide comprehensive analysis data for orthodontics;

[0071] To better annotate image samples for training models, a standard orthodontic model needs to be constructed. Specifically, the following steps are performed: Appearance information of standard upper and lower jaws (herein, appearance information refers to the shape, contour, and dimensions of the upper and lower jaws) is collected; 3D models of the upper and lower jaws are then constructed using a 3D reconstruction algorithm based on this appearance information. Appearance information of standard teeth (herein, appearance information refers to the shape, contour, and dimensions of the standard teeth) is then collected; 3D models of individual teeth are then constructed using a 3D reconstruction algorithm based on this appearance information. The orthodontic model is then constructed by combining all the 3D models of individual teeth with the corresponding 3D models of the upper and lower jaws. This combination is merely a combination of models, not a cohesive model, facilitating subsequent positional transformations of the teeth and upper and lower jaws. The resulting model is a flexible, assembled model. Multiple adjustment points are then applied to the outer contour of the orthodontic model, and these adjustment points are bound to the outer contour of the orthodontic model to create the standard orthodontic model. These adjustment points are referred to as attachment points. When annotating actual orthodontic sample images, the standard orthodontic model can be aligned with the upper and lower jaw and tooth contours in the sample images using these adjustment points. This allows for contour adjustment of the standard orthodontic model, resulting in improved adaptability.

[0072] The 3D models of all single teeth in the standard orthodontic model are separated from the 3D models of the upper and lower jaws (the 3D models of single teeth and the 3D models of the upper and lower jaws are separate models, and there is a model combination state between them for subsequent activity transformation to improve the adaptability of subsequent image annotation); the 3D models of single teeth are sequentially associated in the maxillary dentition and mandibular dentition according to the arrangement order of the standard teeth. The maxillary teeth are usually called the maxillary dentition. Among them, from the middle to the sides are the maxillary central incisor, maxillary lateral incisor, maxillary canine, maxillary first premolar, maxillary second premolar, maxillary first molar, maxillary second molar, maxillary third molar. The mandibular teeth are called the mandibular dentition. The teeth from the middle to the sides are the mandibular central incisor, mandibular lateral incisor, mandibular canine, mandibular first premolar, mandibular second premolar, mandibular first molar, mandibular second molar, and mandibular third molar. Since the arrangement of teeth is orderly, the teeth need to be sorted and associated.

[0073] The standard orthodontic model remains unchanged during use. Here, a standard orthodontic model is copied from the standard orthodontic model as an adjustment model. The adjustment model and the standard orthodontic model are in a superimposed state. The subsequent superposition is set for the adjustment model. The adjustment model also has adjustment points. The adjustment model and the standard orthodontic model are in a superimposed state. The adjustment model is the standard orthodontic model. The adjustment model is used as the transformable subject, and the standard orthodontic model remains unchanged. The lowest point of the root of each standard tooth model in the adjustment model is used as the root point. For each standard tooth model, the root point is used as the center point to define a spatial sphere of a preset diameter. The spatial sphere can be used for tooth angle changes to cover the existing tooth inclination, occlusal relationship, etc., so that it can be used as a database with a large amount of data. The standard tooth model corresponding to each tooth is defined in the spatial sphere with a preset orientation. The preset orientation is the angle range involved in the current tooth inclination. The root point is used as the rotation point to transform the standard tooth model within the preset orientation according to the preset difference angle. The preset difference angle here is a grid on the spherical surface of the spatial sphere. The method comprises the following steps: uniformly dividing the teeth into two groups, and taking the grid on the spherical surface of the space ball corresponding to the preset orientation as the transformation grid, and taking the center point of the grid in the transformation grid as the transformation angle position, so that the preset difference angle is based on the difference angle between the center points of each grid, and obtaining a plurality of transformed standard tooth models; configuring the corresponding model spaces for the plurality of transformed standard tooth models respectively, and storing the plurality of transformed standard tooth models in different model spaces, wherein the model space is a cloud server configured corresponding to each transformed standard tooth model, and the plurality of cloud servers are connected, and the plurality of transformed standard tooth models are overlapped (the superposition process) with the root point as the reference point, and the model spaces of the plurality of transformed standard tooth models are connected to each other, and the plurality of transformed standard tooth models are superimposed through the model space according to the root point to obtain a plurality of tooth superimposed models, which can realize data integration and annotation of many dental problems through the orthodontic data annotation library, can realize large-scale data coverage and accuracy and consistency of image annotation, can quickly annotate the oral-related images of personnel, and improve the efficiency of orthodontic analysis.

[0074] The upper and lower jaws in the standard orthodontic model are superimposed to obtain multiple upper and lower jaw superimposed models. Specifically, the upper and lower jaws are transformed in the adjustment model, such as underbite (mandibular protrusion), underbite (maxillary protrusion), etc., to obtain multiple transformed upper and lower jaw models. Corresponding model spaces are configured for the multiple transformed upper and lower jaw models. The model spaces of the multiple transformed upper and lower jaw models are connected to each other, and the points where the upper and lower jaws are connected are used as reference points for overlap to obtain multiple upper and lower jaw superimposed models.

[0075] Bind multiple tooth superposition models and multiple upper and lower jaw superposition models to corresponding standard orthodontic models respectively, and perform image annotation on multiple tooth superposition models and multiple upper and lower jaw superposition models respectively to obtain an orthodontic data annotation library. Image annotation refers to the process of marking and annotating orthodontic images. Professional orthodontists or experts in related fields classify images into different orthodontic types or mark specific feature information according to the tooth conditions and orthodontic features presented in the images, such as crowded teeth, sparse dentition, deep overbite, open bite and other different orthodontic conditions, to provide accurate labels and classification basis for the training of subsequent deep learning models, so that subsequent models can learn the relationship between image features and corresponding annotations.

[0076] In one embodiment, the labeling unit includes:

[0077] A sample construction unit, used for performing three-dimensional construction of an orthodontic sample image to obtain a sample model;

[0078] An attachment adjustment unit, configured to attach the contour of the sample model via adjustment points on the adjustment model, adjust the contour of the adjustment model, and overlap the adjustment model with the sample model;

[0079] A matching unit, configured to obtain the transformed upper and lower jaw models and the transformed standard tooth model in the adjustment model with the highest degree of overlap corresponding to the sample model;

[0080] an image annotation unit, configured to annotate the images of the transformed upper and lower jaw models and the transformed standard tooth model as image annotations of the sample model;

[0081] It should be noted that after information extraction from the orthodontic sample image, a three-dimensional construction is performed to obtain a sample model. The specific operation can be achieved through X-rays of the upper and lower jaws and dental X-rays to understand the morphology, and through CT scans and images inside the mouth to understand the comprehensive contour information of the upper and lower jaws and teeth, and then the sample model is constructed through the above contour information. After that, the contour of the sample model is attached by adjusting the adjustment points on the model, and the sample model and the adjustment model are initially overlapped. The adjustment points will be attached to the contour of the sample model, and then the adjustment model is adjusted. The adjustment model is overlapped with the standard orthodontic model. The standard orthodontic model does not move, and the background matching changes the adjustment model. The contour of the adjustment model is adjusted, and the adjustment model is overlapped with the sample model. Each model of the adjustment model is an independent combination, so it can be changed according to the adjustment points. The adjustment model is equivalent to multiple independent and single models and is a parts library of model combinations of different forms. There is no need to build a comprehensive database of upper and lower jaws and teeth for multiple situations. Here, the upper and lower jaws and single teeth are respectively regarded as a small part, which is equivalent to a combination of multiple small parts with different forms, different inclinations and positions. The small parts are integrated through a corresponding overall position. There is no need to list a variety of tooth states separately as a whole, which can reduce the complexity of model construction. At the same time, it also has a large amount of data storage. It can be independent and have associated integration, and can obtain the transformed upper and lower jaw models and the transformed standard tooth models in the adjustment model with the highest overlap corresponding to the sample model; then the image annotations of the transformed upper and lower jaw models and the transformed standard tooth models are used as the image annotations of the sample model.

[0082] In one embodiment, the training module includes:

[0083] An information acquisition unit, configured to determine characteristic information of the orthodontic sample image after annotation, wherein the characteristic information is the data size of the orthodontic sample image;

[0084] A selection unit, used to select a deep learning model based on feature information;

[0085] The training unit is used to divide the annotated orthodontic sample images into a training data set and a validation data set, and train the selected deep learning model through the training data set to obtain a trained deep learning model.

[0086] It should be noted that the feature information of the corresponding orthodontic sample images after annotation is determined, wherein the feature information is the data scale of the orthodontic sample images, and the deep learning models are ResNet and VGGNet. ResNet is used when the orthodontic image dataset is large and the image content is complex. The introduction of the residual block structure can effectively solve the problems of gradient vanishing and gradient explosion in deep neural networks, allowing the network to be trained deeper. For example, on a dataset containing a large number of orthodontic X-rays and intraoral photographs taken at different angles and under different lighting conditions, ResNet can learn richer and more abstract features through its deep structure, thereby improving the accuracy of classification. VGGNet is the best choice for orthodontic image datasets that are relatively small. VGGNet has a simple and unified network structure consisting of multiple convolutional layers and pooling layers. On small datasets, it can converge quickly and learn the basic features of the image. The labeled orthodontic sample images are divided into a training dataset and a validation dataset. The selected deep learning model is trained on the training dataset to obtain a trained deep learning model. The trained deep learning model is evaluated on the validation dataset. When the accuracy of the trained deep learning model meets the preset conditions, it is used as the target model. For example, in the orthodontic image classification task, suppose the validation dataset has 1000 images, of which 600 are positive (images requiring orthodontic treatment) and 400 are negative (images not requiring orthodontic treatment). After the model predicts these 1000 images, 500 correctly predict the positive class and 300 correctly predict the negative class, so the accuracy is 80%. If the accuracy is set to 80%, the above model is qualified and serves as the target model, which has a good model determination and training effect.

[0087] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

Claims

1. Orthodontic image classification model determination system based on deep learning, characterized by: include: A construction module is used to construct an orthodontic data annotation library, wherein the orthodontic data annotation library includes multiple tooth superposition models and multiple upper and lower jaw superposition models, and annotate orthodontic sample images according to the orthodontic data annotation library; The training module is connected to the construction module and is used to determine the feature information after annotation, select a deep learning model based on the feature information, and train the selected deep learning model using the annotated orthodontic sample images to obtain a trained deep learning model; The evaluation module is connected to the training module and is used to evaluate the trained deep learning model through the verification data set. When the accuracy of the trained deep learning model meets the preset conditions, it is used as the target model.

2. The orthodontic image classification model determination system based on deep learning according to claim 1, characterized in that: The building blocks include: A construction unit is used to construct a standard orthodontic model, wherein the standard orthodontic model includes upper and lower jaw standard models; The association unit is used to independently establish the association relationship between the teeth in the maxillary dentition and the mandibular dentition for the single tooth models in the standard orthodontic model; A configuration unit is used to configure a space ball for each standard tooth model corresponding to a single tooth in a standard orthodontic model, and to superimpose the single tooth according to the space ball to obtain multiple tooth superimposed models; A position changing unit is used to superimpose the upper and lower jaws in a standard orthodontic model to obtain multiple upper and lower jaw superimposed models; A binding unit is used to bind multiple tooth superposition models and multiple upper and lower jaw superposition models to corresponding standard orthodontic models respectively, and to perform image annotation on the multiple tooth superposition models and the multiple upper and lower jaw superposition models respectively to obtain an orthodontic data annotation library; The annotation unit is used to annotate orthodontic sample images according to the orthodontic data annotation library.

3. The orthodontic image classification model determination system based on deep learning according to claim 2, characterized in that: The building blocks include: The first model building unit is used to collect appearance information of the standard upper and lower jaws and build three-dimensional models of the upper and lower jaws based on the appearance information of the upper and lower jaws; The second model building unit is used to collect appearance information of standard teeth and build a three-dimensional model of each single tooth according to the appearance information of the standard teeth; A combined construction unit is used to combine the three-dimensional models of all individual teeth with the corresponding three-dimensional models of the upper and lower jaws to construct an orthodontic model; The contour binding unit is used to cover multiple adjustment points on the external contour surface of the orthodontic model, bind the adjustment points to the external contour surface of the orthodontic model, and obtain a standard orthodontic model.

4. The orthodontic image classification model determination system based on deep learning according to claim 3, characterized in that: The association unit includes: Independent unit, used to separate the 3D models of all individual teeth in the standard orthodontic model from the 3D models of the upper and lower jaws; The arrangement association unit is used to sequentially associate the three-dimensional models of single teeth in the maxillary dentition and the mandibular dentition according to the arrangement order of the standard teeth.

5. The orthodontic image classification model determination system based on deep learning according to claim 4, characterized in that: The configuration unit includes: The copying unit is used to keep the standard orthodontic model unchanged and copy a standard orthodontic model as an adjustment model, and the adjustment model and the standard orthodontic model are in a superimposed state; A space configuration unit is configured to configure a space sphere corresponding to each standard tooth model in the adjustment model, and perform angle transformation on the standard tooth model of a single tooth in the space sphere based on a root point according to a preset orientation to obtain multiple transformed standard tooth models; The transformation unit is used to configure the corresponding model space of multiple transformed standard tooth models respectively, connect the model spaces of multiple transformed standard tooth models to each other, and superimpose the multiple transformed standard tooth models through the model space according to the root point to obtain multiple tooth superimposed models.

6. The orthodontic image classification model determination system based on deep learning according to claim 5, characterized in that: The space configuration unit includes: a delineation unit, configured to use the lowest point of the tooth root of each standard tooth model in the adjustment model as a root point, and to delineate a spatial sphere of a preset diameter using the root point as a center point for each standard tooth model; The transformation unit is used to define a preset orientation of the standard tooth model corresponding to each tooth in the space sphere, and use the root point as the rotation point to transform the standard tooth model within the preset orientation according to a preset difference angle to obtain multiple transformed standard tooth models.

7. The orthodontic image classification model determination system based on deep learning according to claim 6, characterized in that: The marking unit includes: A sample construction unit, used for performing three-dimensional construction of an orthodontic sample image to obtain a sample model; An attachment adjustment unit, configured to attach the contour of the sample model via adjustment points on the adjustment model, adjust the contour of the adjustment model, and overlap the adjustment model with the sample model; A matching unit, configured to obtain the transformed upper and lower jaw models and the transformed standard tooth model in the adjustment model with the highest degree of overlap corresponding to the sample model; The image annotation unit is used to use the image annotations of the transformed upper and lower jaw models and the transformed standard tooth model as the image annotations of the sample model.

8. The orthodontic image classification model determination system based on deep learning according to claim 1, characterized in that: The training module includes: An information acquisition unit, configured to determine characteristic information of the orthodontic sample image after annotation, wherein the characteristic information is the data size of the orthodontic sample image; A selection unit, used to select a deep learning model based on feature information; The training unit is used to divide the annotated orthodontic sample images into a training data set and a validation data set, and train the selected deep learning model through the training data set to obtain a trained deep learning model.

Images