Method for constructing three-dimensional digital model of teeth, generating treatment plan and preparing appliance

Abstract

This invention discloses a method for constructing a three-dimensional digital model of teeth, comprising the following steps: acquiring CBCT image data of the patient's current oral cavity and a three-dimensional digital model of teeth representing the patient's current dentition; segmenting the CBCT image data using a pre-trained segmentation model of abnormal parts of teeth and oral cavity to obtain a CBCT virtual model of the patient's current dentition and obtain abnormal part identification results; registering the CBCT virtual model with the identified abnormal parts with the three-dimensional digital model of teeth, mapping the abnormal part identification results in the CBCT virtual model to the three-dimensional digital model of teeth, and obtaining a three-dimensional digital model of teeth with abnormal part information. This invention utilizes a deep neural network to obtain a three-dimensional digital model of teeth with abnormal part information. Correspondingly, this invention also provides a method for generating orthodontic treatment plans and a method for preparing shell-shaped orthodontic appliances.

Description

Technical Field

[0001] This application relates to the field of orthodontic technology, and in particular to a method for constructing a three-dimensional digital model of teeth, generating a treatment plan, and preparing orthodontic appliances. Background Technology

[0002] Currently, shell-shaped orthodontic appliances based on polymer materials are becoming increasingly popular due to their advantages such as aesthetics, convenience, and ease of cleaning. For orthodontic treatment of a single jaw (maxillary or mandibular) teeth, a set of shell-shaped appliances typically includes a dozen or even dozens of successive shell-shaped appliances used to gradually reposition the teeth from an initial layout to a target layout. This includes N successive intermediate layouts, from a first intermediate layout to a final intermediate layout, between the initial and target layouts.

[0003] The aforementioned series of shell-shaped orthodontic appliances are made based on a three-dimensional digital model representing the series of successive tooth layouts, which are commonly referred to as the orthodontic treatment plan for the dentition. This plan specifies the sequence of tooth movement, including the order in which different teeth are moved.

[0004] Orthodontic treatment plans are typically based on the patient's initial and target tooth layouts. A series of intermediate digital target dental models are derived from the initial and target 3D digital models of the teeth, representing the initial and target tooth layouts, respectively. These intermediate digital target dental models represent a series of orthodontic states in which the patient's teeth are progressively moved from the initial tooth layout to the target tooth layout. Then, based on these intermediate digital target dental models, physical dental models are 3D printed and thermoforming is used to obtain a series of shell-shaped appliances. Alternatively, a series of shell-shaped dental appliances can be directly fabricated using additive manufacturing processes based on these intermediate digital target dental models.

[0005] However, in many cases, patients awaiting orthodontic treatment often have pre-existing oral or dental conditions, such as cavities, periodontitis, and gingivitis. For such patients, the orthodontic treatment plan first requires identifying the location and type of their oral disease before determining the appropriate treatment plan. Currently, the diagnosis of the location and type of oral disease relies on the doctor's experience, and the results depend heavily on that experience. For these patients, the current orthodontic treatment plan generally involves treating the oral disease first, followed by orthodontic treatment. This is feasible for patients who do not have a time limit for treatment, but for patients who expect better results quickly, treating the oral disease first and then starting orthodontic treatment will delay the achievement of their treatment goals, making the treatment plan unacceptable. Summary of the Invention

[0006] The main objective of this invention is to propose a method for constructing a three-dimensional digital model of teeth, generating orthodontic plans, and preparing orthodontic appliances. This method utilizes a deep neural network model based on the patient's current oral CBCT image data to detect and identify abnormal areas in the oral cavity, and maps the identification results onto the three-dimensional digital model of teeth obtained from an oral scan. This allows for the automatic generation of an orthodontic plan for the patient with abnormal oral areas based on the three-dimensional digital model of teeth containing information about abnormal areas.

[0007] To achieve the above objectives, embodiments of this application provide a method for constructing a three-dimensional digital model of teeth, comprising the following steps:

[0008] Acquire CBCT imaging data of the patient's current oral cavity and a three-dimensional digital model of the teeth representing the patient's current dentition;

[0009] The CBCT image data is segmented using a pre-trained segmentation model of abnormal parts of teeth and oral cavity to obtain a CBCT virtual model of the patient's current dentition, and the abnormal part identification results of the CBCT virtual model are obtained. The abnormal part identification results include at least the abnormality type and abnormality location information.

[0010] The CBCT virtual model with the identified abnormal parts is registered with the three-dimensional digital model of the tooth. Based on the registration result, the abnormal part identification result in the CBCT virtual model is mapped to the three-dimensional digital model of the tooth to obtain a three-dimensional digital model of the tooth with abnormal part information. The abnormal part information includes at least the abnormal type and abnormal location information.

[0011] Optionally, the segmentation model for teeth and oral cavity abnormalities is trained from multiple pre-acquired training samples, and the training includes:

[0012] Based on historical data from multiple patients, several sample CBCT images were selected for different categories of dental abnormalities and labeled with features to construct a sample CBCT image dataset. The label feature annotation includes target localization annotation and abnormality classification annotation.

[0013] Image features of training samples in the CBCT image dataset are extracted using a feature extraction model.

[0014] The image features are identified using a feature recognition model to obtain the target recognition result of the training sample. The target result includes the tooth classification result and the abnormal part recognition result.

[0015] Optionally, the segmentation model for teeth and oral cavity abnormalities is a neural network model for two-dimensional image segmentation. The training sample is input into the segmentation model for teeth and oral cavity abnormalities as a two-dimensional slice sequence to obtain the recognition result of each two-dimensional slice. The recognition results of all two-dimensional slices are stacked to obtain the target recognition result of the training sample.

[0016] Optionally, the segmentation model for teeth and oral cavity abnormalities is a neural network model for three-dimensional image segmentation, and the training sample is obtained by inputting a three-dimensional image into the segmentation model for teeth and oral cavity abnormalities.

[0017] Optionally, the abnormality type includes at least one of periodontitis, pulpitis, gingivitis, dental caries, alveolar osteitis, and periapical periodontitis.

[0018] Optionally, the registration of the CBCT virtual model with identified abnormal areas to the three-dimensional digital model of the teeth, and the mapping of the abnormal area identification results in the CBCT virtual model to the three-dimensional digital model of the teeth based on the registration results, includes:

[0019] The point cloud data of the CBCT virtual model and the three-dimensional digital model of the teeth are obtained as the source point cloud and the target point cloud, and the principal axis directions of the two sets of point clouds are extracted.

[0020] Calculate the rotation matrix between the two sets of point clouds based on the principal axis directions of the two sets of point clouds;

[0021] Based on the rotation matrix and translation vector, the affine transformation matrix between the two sets of point clouds is obtained;

[0022] The affine transformation matrix is ​​used to map the abnormal area information of the CBCT virtual model to the three-dimensional digital model of the teeth.

[0023] To achieve the above objectives, the present invention also provides a method for generating a dental orthodontic treatment plan, comprising the following steps:

[0024] The three-dimensional digital model of teeth of the current patient is obtained according to the aforementioned method of constructing a three-dimensional digital model of teeth, which is used as the first digital dental model representing the layout of the first dentition of the patient. The information on the abnormal parts of the teeth includes at least the abnormal type and abnormal location information of the current patient.

[0025] The treatment and correction phases of the current patient's treatment plan are determined based on the type of abnormality.

[0026] Obtain a second digital tooth model representing the current patient's second dentition layout;

[0027] Based on the determined treatment and orthodontic complex stage and the first and second digital tooth models, a series of target digital tooth models are obtained. The series of target digital tooth models represent a series of orthodontic states in which the patient's teeth are progressively moved from the first dentition layout to the second dentition layout. The series of target digital tooth models correspond to the treatment and orthodontic complex stage and the subsequent individual orthodontic stages. Each target digital tooth model corresponding to the treatment and orthodontic complex stage has corresponding abnormality information.

[0028] Based on the information about the abnormal locations, corresponding digital models of drug storage structures are generated on the target digital tooth models in the combined treatment and orthodontic stage.

[0029] Optionally, the single-step design movement amount of each movement mode of each target digital tooth model in the treatment and orthodontic complex stage is less than the single-step design movement amount of the corresponding movement mode of each target digital tooth model in the individual orthodontic stage.

[0030] Optionally, the single-step design movement amount of each movement mode of the target digital tooth model in the treatment and orthodontic complex stage increases as the treatment plan progresses.

[0031] Optionally, the digital model of the drug storage structure covers at least half of the area of ​​the corresponding abnormal part.

[0032] To achieve the above objectives, the present invention also provides a method for preparing a shell-shaped orthodontic appliance, the method comprising:

[0033] Based on the aforementioned method for generating orthodontic treatment plans, a series of target digital tooth models representing the gradual movement of a patient's dentition from a first dentition layout to a second dentition layout are obtained. The series of target digital tooth models correspond to the treatment-orthodontic composite stage and subsequent individual orthodontic stages. Each target digital tooth model in the treatment-orthodontic composite stage has information on abnormal sites and a digital model of a drug storage structure corresponding to the information on abnormal sites.

[0034] A series of shell-shaped orthodontic appliances are prepared based on the target digital tooth models, wherein each shell-shaped orthodontic appliance corresponding to the treatment and orthodontic composite stage has a drug storage structure on its shell corresponding to the abnormal site for storing drugs.

[0035] Based on the current patient's abnormality type, corresponding medications are added to the medication storage structures of each shell-shaped orthodontic appliance corresponding to the treatment-orthodontic complex stage, so as to provide medication to the abnormal site when the appliance is worn on the patient's dentition during the treatment-orthodontic complex stage.

[0036] Optionally, the elasticity of the shell of each shell-shaped orthodontic appliance in the treatment-treatment complex stage is greater than the elasticity of the shell of each shell-shaped orthodontic appliance in the individual orthodontic stage.

[0037] Optionally, the elasticity of the shell of each shell-shaped orthodontic appliance in the treatment and orthodontic complex phase gradually decreases with the order of treatment.

[0038] To achieve the above objectives, the present invention also provides an electronic device, comprising:

[0039] At least one processor; and,

[0040] A memory communicatively connected to the at least one processor; wherein,

[0041] The memory stores instructions that can be executed by the at least one processor, which enable the at least one processor to perform the above-described method for constructing a three-dimensional digital model of teeth or the above-described method for generating orthodontic treatment plans.

[0042] Compared with existing technologies, the present invention provides a method for constructing a three-dimensional digital model of teeth, generating a treatment plan, and preparing orthodontic appliances, which has the following beneficial effects:

[0043] 1. This invention can use a trained segmentation model of teeth and oral cavity abnormalities to identify oral cavity abnormalities in CBCT image data, and map the identification results of abnormalities onto the three-dimensional digital model of teeth obtained from oral scanning. The identification efficiency is high and does not rely on the experience of professionals such as doctors or designers. The obtained three-dimensional digital model of teeth contains information on abnormalities, so the orthodontic plan can be automatically planned based on this three-dimensional digital model of teeth.

[0044] 2. This invention can determine the target digital tooth model corresponding to the treatment and orthodontic complex stage based on the abnormality type of the identified abnormal site, and make each target digital tooth model in the treatment and orthodontic complex stage have corresponding abnormal site information, and generate a corresponding drug storage structure digital model based on the abnormal site information, so that the shell-shaped orthodontic appliance prepared accordingly has a corresponding drug storage structure, thereby storing drugs in the treatment and orthodontic complex stage to enable patients to receive treatment and orthodontic treatment at the same time, improving the efficiency of the current patient's orthodontic treatment.

[0045] 3. This invention can directly prepare a shell-shaped orthodontic appliance with a drug storage structure based on the digital model of the drug storage structure on the digital tooth model of each target in the treatment-orthodontic complex stage. By adding drugs targeting the abnormal type inside the drug storage structure, the abnormal area can be treated at the same time as the teeth are being corrected when the patient wears the shell-shaped orthodontic appliance in the treatment-orthodontic complex stage. The simultaneous correction and treatment improve the efficiency of the patient's orthodontic treatment. Attached Figure Description

[0046] One or more embodiments are illustrated by way of example with reference to the accompanying drawings, and these illustrative descriptions do not constitute a limitation on the embodiments.

[0047] Figure 1 This is a flowchart of a method for constructing a three-dimensional digital model of teeth according to the first embodiment of the present invention;

[0048] Figure 2 This is a schematic diagram of a CBCT image in the first embodiment of the present invention;

[0049] Figure 3 This is a schematic diagram of the annotation of CBCT images in the first embodiment of the present invention;

[0050] Figure 4 This is a schematic diagram of a two-dimensional slice of a CBCT image in the first embodiment of the present invention;

[0051] Figure 5 This is a schematic diagram of the model registration process in one embodiment of the first embodiment of the present invention;

[0052] Figure 6 This is a schematic diagram of a CBCT virtual model used for registration in one embodiment of the first embodiment of the present invention;

[0053] Figure 7 This is a schematic diagram of a three-dimensional digital model of teeth used for registration in one embodiment of the first embodiment of the present invention;

[0054] Figure 8 This is a flowchart of a method for generating a treatment plan according to a second embodiment of the present invention;

[0055] Figure 9 This is a flowchart of the preparation method of the shell-shaped orthodontic appliance according to the third embodiment of the present invention;

[0056] Figure 10 This is a schematic diagram of a shell-shaped orthodontic appliance with a drug storage structure according to one embodiment of the third embodiment of the present invention;

[0057] Figure 11 This is a schematic diagram of a shell-shaped orthodontic appliance with a drug storage structure, according to another embodiment of the third embodiment of the present invention.

[0058] Figure 12 This is a schematic diagram of a shell-shaped orthodontic appliance with a drug storage structure, which is another embodiment of the third embodiment of the present invention. Detailed Implementation

[0059] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the various embodiments of this application will be described in detail below with reference to the accompanying drawings. However, those skilled in the art will understand that many technical details have been provided in the various embodiments of this application to help readers better understand this application. However, the technical solutions claimed in this application can be implemented even without these technical details and various changes and modifications based on the following embodiments. The division of the various embodiments below is for the convenience of description and should not constitute any limitation on the specific implementation of this application. The various embodiments can be combined with and referenced by each other without contradiction.

[0060] To keep the drawings concise, only the parts relevant to the invention are shown schematically in each figure, and they do not represent the actual structure of the product. Furthermore, for ease of understanding, in some figures, only one of components with the same structure or function is shown schematically, or only one is labeled. In this document, "one" can mean not only "only one" but also "more than one".

[0061] The inventors of this application have discovered that, currently, orthodontic treatment for patients with oral diseases generally involves doctors or designers assessing the patient's oral abnormalities based on experience, determining the type of abnormality, and then confirming the treatment and orthodontic plan. The orthodontic treatment plan typically involves treating the oral disease first, followed by orthodontic treatment. It is evident that the current identification of oral abnormalities relies heavily on the professional experience of doctors or designers, resulting in low efficiency and a high risk of misjudgment due to personal experience. Therefore, the orthodontic treatment plan of treating the oral disease first and then proceeding with orthodontic treatment may be feasible for some patients who do not have a time limit for orthodontic treatment. However, for patients who expect to achieve better orthodontic results as quickly as possible, this plan will delay the achievement of their orthodontic goals and fail to meet their objectives.

[0062] In some embodiments, this invention proposes a method for constructing a three-dimensional digital model of teeth. This method involves acquiring CBCT image data of the patient's current oral cavity, segmenting the CBCT image data using a pre-trained segmentation model for teeth and oral cavity abnormalities, obtaining a CBCT virtual model that identifies abnormal areas, and then registering this CBCT virtual model with a three-dimensional digital model of teeth obtained from an intraoral scan. Based on the registration result, the abnormal area information in the CBCT virtual model is mapped to the three-dimensional digital model of teeth obtained from the intraoral scan, thereby obtaining a three-dimensional digital model of teeth with abnormal area information. This invention utilizes a deep neural network based on CBCT image data to identify oral abnormalities, achieving high recognition efficiency without relying on the experience of professionals such as doctors or designers. The obtained three-dimensional digital model of teeth with abnormal area information allows for automatic planning of orthodontic treatment plans.

[0063] In some embodiments, based on obtaining a three-dimensional digital model of teeth with information on abnormal locations, this invention also proposes a method for generating orthodontic treatment plans based on this three-dimensional digital model of teeth with information on abnormal locations. First, a first digital tooth model representing the patient's first dentition layout and a second digital tooth model representing the current patient's second dentition layout are determined. Then, the treatment-orthodontic composite stage of the current patient's treatment plan is determined according to the abnormality type in the abnormal location information. This treatment-orthodontic composite stage is used for simultaneous treatment and orthodontic treatment. Then, based on the determined treatment-orthodontic composite stage and the first and second digital tooth models, a series of target digital tooth models are obtained, and the corresponding treatment-orthodontic composite stages are... Each target digital tooth model contains corresponding abnormality information. Finally, in the treatment-orthodontic composite stage, a corresponding digital model of the drug storage structure is generated on each target digital tooth model corresponding to the abnormality information. It can be seen that the present invention realizes the automated generation of orthodontic plans based on the three-dimensional digital tooth models with abnormality information. This ensures that each target digital tooth model in the treatment-orthodontic composite stage, which is used for simultaneous treatment and orthodontic treatment, has corresponding abnormality information, and a corresponding digital model of the drug storage structure is generated based on the abnormality information. Accordingly, the prepared shell-shaped orthodontic appliance has a corresponding drug storage structure, which can store drugs in the treatment-orthodontic composite stage to enable patients to receive treatment while undergoing orthodontic treatment, thereby improving the efficiency of orthodontic treatment.

[0064] In some embodiments, after generating a dental orthodontic treatment plan, the present invention also proposes a method for preparing shell-shaped orthodontic appliances. A series of shell-shaped orthodontic appliances can be prepared based on the above-mentioned series of target digital tooth models. The corresponding digital model of the corresponding drug storage structure on each shell-shaped orthodontic appliance corresponding to the treatment-orthodontic composite stage forms the corresponding drug storage structure. Then, according to the current patient's abnormality type, the corresponding drug is added to the drug storage structure of each shell-shaped orthodontic appliance so as to provide drug treatment to the corresponding abnormal area when each shell-shaped orthodontic appliance is worn on the patient's dentition during the treatment-orthodontic composite stage, thereby achieving the purpose of simultaneous treatment of oral diseases and orthodontics.

[0065] The implementation details of the shell-shaped dental instrument described in this application will be specifically described below with reference to specific embodiments. The following implementation details are provided for ease of understanding only and are not necessary for implementing this solution.

[0066] The first embodiment of the present invention provides a method for constructing a three-dimensional digital model of teeth, the process of which is as follows: Figure 1 As shown, the method for constructing the three-dimensional digital model of the teeth includes the following steps:

[0067] Step S101: Obtain CBCT image data of the patient's current oral cavity and a three-dimensional digital model of the teeth representing the patient's current dentition.

[0068] Step S102: The CBCT image data is segmented using a pre-trained segmentation model of abnormal parts of teeth and oral cavity to obtain a CBCT virtual model of the patient's current dental arch, and the abnormal part identification result of the CBCT virtual model is obtained. The abnormal part identification result includes at least the abnormal type and abnormal location information.

[0069] Step S103: Register the CBCT virtual model with the identified abnormal parts with the three-dimensional digital model of the teeth. Based on the registration result, map the abnormal part identification result in the CBCT virtual model to the three-dimensional digital model of the teeth to obtain a three-dimensional digital model of the teeth with abnormal part information. The abnormal part information includes at least the abnormal type and abnormal location information.

[0070] As can be seen, this embodiment obtains a segmentation model of abnormal parts of teeth and oral cavity by training a deep neural network model, and then uses this segmentation model to identify abnormal parts of oral cavity in CBCT image data. The identification results are then mapped onto the three-dimensional digital model of teeth obtained from the oral scan. The identification efficiency is high and does not rely on the experience of professionals such as doctors or designers. The obtained three-dimensional digital model of teeth contains information on abnormal parts, which can be used to automatically plan orthodontic treatment plans.

[0071] The following is a detailed description of each step in the first embodiment of the present invention:

[0072] In step S101, firstly, CBCT image data of the patient's oral cavity is acquired, also known as oral CBCT images. Oral CBCT images are obtained by using cone-beam CT, performing a circular digital projection around the patient's oral cavity, and then reconstructing the data obtained from the intersection of multiple digital projections around the projection body in a computer to obtain a three-dimensional image. This process also determines the overall tooth region image within the oral cavity, such as... Figure 1 The image shown is a schematic diagram of an oral CBCT image obtained in an example.

[0073] A three-dimensional digital model of teeth representing a patient's current dentition refers to a digital model of the patient's dentition to be treated, obtained through intraoral scanning. In some embodiments, this can be achieved by directly scanning the patient's jaw to obtain a three-dimensional digital model of the current dentition. In another embodiment, this can be achieved by scanning a physical model of the patient's jaw, such as a plaster cast. In yet another embodiment, this can be achieved by scanning a bite impression of the patient's jaw to obtain a three-dimensional digital model of the current dentition. In this embodiment, the three-dimensional digital model of teeth is a mesh digital model obtained by segmenting the teeth from the intraoral scan results.

[0074] In step S102, for the oral CBCT image, a pre-trained segmentation model of teeth and oral abnormalities can be used to segment the oral CBCT image. The segmentation result yields a CBCT virtual model of the patient's current dentition and also obtains abnormality identification results. The abnormality identification results include at least the abnormality type and abnormality location information, that is, a CBCT virtual model with abnormality information is obtained.

[0075] In some examples, a preprocessing step for the CBCT image may be included before step S102. This preprocessing includes adjusting the window width and window level of the CBCT image and normalizing it to make teeth, alveolar bone, and abnormal areas in the CBCT image clearer. The window width refers to the range of CT values ​​displayed on the CBCT image, which directly affects the image's contrast. A narrower window width results in a smaller range of CT values ​​and higher contrast, which is beneficial for observing details of specific tissues. A wider window width results in a larger range of CT values ​​and lower contrast, suitable for observing tissues with significant density differences. The window level refers to the average of the upper and lower limits of the window width, affecting the brightness of the CBCT image. A lower window level results in higher image brightness, while a higher window level results in lower image brightness. By adjusting the window width and window level of the CBCT image, the accuracy of the segmentation results of the segmentation model for teeth and oral abnormalities can be improved.

[0076] In this embodiment, the segmentation of abnormal areas and teeth in CBCT images is achieved based on a pre-trained segmentation model for teeth and oral cavity abnormalities. This model is a neural network model, trained from multiple pre-acquired training samples. First, sample CBCT images are selected based on historical patient CBCT image data and labeled to construct a sample CBCT image dataset. In this embodiment, a specific number of sample CBCT images need to be obtained from historical data according to the type of oral disease. Each CBCT image is labeled with features. That is, for each type of oral disease, a specific number of sample CBCT images needs to be obtained. The specific number is set according to the actual situation. If the number of a certain type in the historical data does not reach the specific number, the selection range is expanded to reach the specific number. If it still does not reach the specific number, it is labeled, and additional samples are collected. Of course, for historical patient CBCT image data, CBCT images can also be preprocessed before labeling and constructing the sample CBCT image dataset.

[0077] Each training sample in the CBCT image dataset is labeled with the target region and corresponding category for each tooth body, alveolar bone, and abnormal site. Figure 2 This diagram illustrates a CBCT image of a patient's oral cavity. In this embodiment, different colors can be used to label the target areas and corresponding types of the tooth body, alveolar bone, and abnormal sites in each training sample, such as... Figure 3 As shown, the abnormal site types include at least one of periodontitis, pulpitis, gingivitis, dental caries, alveolar osteitis, and periapical periodontitis. Therefore, the categories labeled in each training sample must include at least the tooth category, different types of abnormal sites, and the background (non-abnormal area). In some specific examples, the total number of categories labeled in each training sample is 41, including at least the background, maxilla, mandible, 4*8 teeth, and at least 6 types of tooth abnormalities. For the 4*8 teeth, FDI can be used to distinguish their categories, for example, different regions of permanent teeth are represented by quadrants 1-4, such as 11 representing tooth number 1 in the first quadrant of the maxilla. Of course, the FDI number of the teeth can not be used to label the tooth categories in the training samples; other identifiers can be used to label each tooth category, as long as it is possible to distinguish teeth in different positions based on the identifier. In this embodiment, the labeling of the training samples can be done manually, that is, the categories and target areas of the teeth and other tissues and abnormal sites in the training samples are manually identified and labeled.

[0078] After constructing the sample CBCT image dataset, 70% of the dataset can be selected as the training set and 30% as the test set to train the segmentation model for teeth and oral cavity abnormalities.

[0079] In this embodiment, the constructed segmentation model for abnormal parts of teeth and oral cavity includes a feature extraction model and a feature recognition model. The feature extraction model is used to extract image features from the training samples in the sample CBCT image dataset, and the feature recognition model is used to recognize the image features to obtain the target recognition result of the training samples. The target result includes the tooth classification result and the abnormal part recognition result. The tooth classification result is, for example, the target localization area and its location of each tooth and the corresponding tooth type. The abnormal part recognition result is the target localization area and its location of the abnormal part and the corresponding abnormal type.

[0080] Since CBCT images are composed of stacked multi-layer images of the oral cavity acquired using CBCT technology, one method for segmentation is to separately identify each layer (also known as a CBCT 2D slice image) using a deep neural network model for 2D image recognition, and then stack the recognition results. This can be any trained deep neural network model for 2D image recognition, such as Unet, Unet++, TransUNet, Vnet, etc., and this embodiment is not limited to any particular model. In some implementations, the input to the segmentation model for teeth and oral cavity abnormalities is a sequence of 2D slices from CBCT images, such as... Figure 4 The diagram shows a schematic of one of the two-dimensional slices. The feature extraction module includes an encoding module, comprising at least three encoders connected in series: a first encoder, a second encoder, and a third encoder. The first encoder extracts the edge information of the target in the sample CBCT image data. The second encoder extracts the texture and shape information of the target based on the edge information. The third encoder obtains the contour features of the target based on the texture and shape information. The feature recognition model includes a decoding module, comprising at least a first decoder, a second decoder, and a third decoder connected in series. The first decoder decodes the contour features of the target to obtain a first decoding result. The second decoder upsamples the first decoding result and concatenates it with the texture and shape information of the target to obtain a second decoding result. The third decoder upsamples the second decoding result and concatenates it with the edge information of the target to obtain a third decoding result. The third decoding result is processed by an activation function to obtain the class probability of each pixel.

[0081] Specifically, the two-dimensional slice sequence of the sample CBCT image is input into the tooth and oral cavity abnormality recognition model. The encoding module extracts feature information from the two-dimensional slice data through convolution and pooling operations. The decoding module restores the feature information to the same size as the input data through deconvolution and upsampling operations. The first encoder extracts low-level features through convolution and pooling to obtain the edge information of the target. The second encoder obtains the texture and shape features of the target through convolution and pooling. The third encoder extracts high-level contour features through convolution and pooling. The first decoder upsamples the high-level contour features extracted by the third encoder and then combines them with the texture and shape features extracted by the second encoder. The image is composed of concatenated features and then subjected to a series of convolutional layers for feature extraction and upsampling to obtain an intermediate image of the same size as the input image. This intermediate image is the third decoding result. The second decoder upsamples the third decoder and then concatenates it with the texture and shape features. Finally, a series of convolutional layers are used for feature extraction and upsampling to obtain the second decoding result. The third decoder upsamples the second decoding result and then concatenates it with the edge information. Finally, a series of convolutional layers are used for feature extraction and upsampling to obtain the third decoding result. The third decoding result is then activated by the Softmax activation function to obtain the probability of each pixel corresponding to the category.

[0082] In some specific implementations, the tooth and oral cavity abnormality identification model can continuously update the model's weight parameters through backpropagation to minimize the loss function.

[0083] This embodiment uses the above-mentioned test set to test the tooth and oral cavity abnormality recognition model to obtain performance indicators such as the model's segmentation accuracy, so as to evaluate the model's generalization ability.

[0084] This embodiment utilizes a deep neural network model for 2D image recognition to segment each two-dimensional slice of the CBCT image's two-dimensional slice sequence. Then, all segmentation results are stacked to obtain a three-dimensional segmentation result, thus obtaining a CBCT virtual model of the patient's current dental arch with abnormal site identification results.

[0085] In other embodiments, CBCT images composed of multiple layers of two-dimensional slices are used as three-dimensional images and can be directly identified using neural network models for 3D image recognition. Similarly, this embodiment can also use any trained deep neural network model for 3D image recognition, such as 3D-Unet, 3D-Vnet, etc., and this embodiment is not limited. In some specific embodiments, the feature extraction module of the tooth and oral cavity abnormality recognition model includes an encoding module, and the feature recognition module includes a decoding module and a sampling fusion module. The encoding module includes a first convolutional layer and a downsampling layer, and the decoding module includes a second convolutional layer and an upsampling layer; the encoding module is used to obtain a first feature map of the sample CBCT image; the decoding module is used to process the first feature map to obtain a second feature map of the sample CBCT image; both the first feature map and the second feature map contain multiple image scales; the sampling fusion module is used to perform a first feature fusion process on the first feature map, and then link the result of the first feature fusion process to the second feature map for a second feature fusion process.

[0086] Specifically, the first convolutional layer of the encoding module is used for feature extraction, and the resulting feature map is denoted as the first feature map. The first convolutional layer can consist of a set of filters, which can be considered as mathematical matrices. Optionally, Gaussian filters can be used. In an optional embodiment, a batch normalization layer and an activation layer can be sequentially connected after the first convolutional layer of the encoding module to perform batch normalization and activation processing on the feature extraction results of the sample CBCT images. Specifically, the batch normalization layer is used to perform batch normalization processing, that is, when the model is trained by stochastic gradient descent, the corresponding response is normalized so that the mean of the output result is 0 and the variance is 1. This can speed up the convergence of the model and make the trained deep neural network model more stable. The activation layer introduces nonlinear features for activation processing by using appropriate activation functions, enabling the model to cope with learning or simulating more complex data and improving the model's learning ability. The activation layer can use Sigmoid function, Logistic function, ReLU linear correction unit, etc. as activation functions.

[0087] The downsampling layer of the encoding module, also called the pooling layer, is used to transform the image scale of the first feature map (generally, the image scale (length and width) is reduced with each downsampling, and the reduction ratio can be preset, such as half, one-third, one-quarter, etc.), thus obtaining a first feature map containing multiple image scales. Common downsampling methods are max pooling or mean pooling. After being downsampled, the first feature map is input into the first convolutional layer, which can expand the receptive field, extract low-resolution information from the sample CBCT image data, and thus provide contextual information about the location of oral diseases in the entire sample CBCT image (i.e., the input and output of a certain feature map obtained in the encoding process are related to other feature maps). By extracting the first feature map at multiple image scales, features reflecting the relationship between the target and its environment can be obtained, which is helpful for class determination in classification problems.

[0088] The upsampling layer of the decoding module is used to recover the image scale of the feature map of the sample CBCT image (generally, the image scale (length and width) is expanded with each upsampling, and the expansion factor can be preset, such as 2x, 3x, 4x, etc.), to obtain the second feature map. Common upsampling methods include linear interpolation or deconvolution. The second convolutional layer of the decoding module can be used to extract high-resolution information from the second feature map, providing accurate localization and segmentation basis, making the edge information of the target segmentation result more refined. Similar to the first convolutional layer, the second convolutional layer can be implemented using filters.

[0089] The downsampling layer of the encoding module and the upsampling layer of the decoding module can be set accordingly so that the number of image scale layers of the first feature map and the second feature map obtained in the oral cavity abnormality recognition model are corresponding. Specifically, the number of downsampling layers and upsampling layers can be no less than 2, so that the number of image scale layers of the first feature map and the second feature map obtained in the oral cavity abnormality recognition model is at least 3.

[0090] The sampling and fusion module performs a first feature fusion process on the first feature map and links the result of the first feature fusion process to the second feature map for a second feature fusion process. Finally, the result of the second feature fusion process is convolved and mapped to the model's output. Specifically, the second feature fusion process involves fusing the result of the first feature fusion process at the current layer's image scale with the second feature map at the current layer's image scale, which is upsampled and output by the decoding module. Specifically, the second feature fusion process can be performed by concatenating the feature dimensions of the feature maps.

[0091] In this embodiment, the encoding module can obtain multi-layer feature information of the sample CBCT image after multiple downsamplings, and the sampling fusion module can perform flexible fusion operations on this information, so that the model can make full use of these features and improve the accuracy of the segmentation results without increasing the model's downsampling depth. The result of the first feature fusion processing is linked to the second feature map for second feature fusion processing. The high-resolution information extracted by the decoding module can provide more refined features for segmentation, ensuring the accuracy of the segmentation results.

[0092] The trained segmentation model for abnormal parts of teeth and oral cavity can be used to directly segment the current patient's 3D CBCT image, thereby obtaining a CBCT virtual model with abnormal part information. The abnormal part information includes at least the location information of the abnormal part and the type of abnormality, such as which type of disease it is: periodontitis, pulpitis, gingivitis, dental caries, alveolar osteitis, or periapical periodontitis.

[0093] In step S103, the CBCT virtual model with the identified abnormal parts is registered with the three-dimensional digital model of the teeth to obtain an affine transformation matrix. Based on the affine transformation matrix, the abnormal part identification results in the CBCT virtual model are mapped to the three-dimensional digital model of the teeth obtained by intraoral scanning.

[0094] Because dental models obtained through intraoral scanning provide doctors with relatively accurate oral structural data, enabling more precise orthodontic treatment, these 3D digital dental models are commonly used in orthodontic treatment planning. However, when a patient has abnormal areas in their mouth, this information cannot be obtained from the intraoral 3D digital dental model, making it impossible to provide a personalized orthodontic treatment plan based on this information. Therefore, after obtaining a CBCT virtual model that identifies abnormal areas, it is necessary to map the abnormal area information from this CBCT virtual model onto the 3D digital dental model obtained through intraoral scanning.

[0095] In this embodiment, in order to map the abnormal part information on the CBCT virtual model to the three-dimensional digital model of the teeth obtained by oral scanning, the CBCT virtual model and the three-dimensional digital model of the teeth obtained by oral scanning can be registered to obtain the affine transformation matrix between the two, and then the abnormal part information on the CBCT virtual model can be mapped to the three-dimensional digital model of the teeth obtained by oral scanning based on the affine transformation matrix.

[0096] In some specific implementation methods, such as Figure 5 As shown, the registration of the CBCT virtual model with the identified abnormal area and the three-dimensional digital model of the tooth to obtain the affine transformation matrix includes:

[0097] Step S103a: Obtain the point cloud data of the CBCT virtual model and the three-dimensional digital model of the teeth as the source point cloud and the target point cloud, respectively, and extract the main axis direction of the two sets of point clouds.

[0098] In some examples, the CBCT virtual model is first obtained (e.g. Figure 6 Using the point cloud data (as shown) as the source point cloud, a three-dimensional digital model of the teeth obtained from an oral scan (such as...) is acquired. Figure 7 The point cloud data shown is used as the target point cloud, and then the principal axis directions of the source point cloud P and the target point cloud Q are extracted using the principal component analysis (PCA) method. Specifically, the centroids of the source point cloud P and the target point cloud Q are first calculated according to the following formulas:

[0099]

[0100] Where P i Q i Let P and Q represent individual point clouds, respectively. and Let P and Q represent the centroids of the source point cloud and the target point cloud, respectively.

[0101] Calculate the covariance matrices of the source point cloud P and the target point cloud Q based on their centroids:

[0102]

[0103] Among them, M P With M Q Let P and Q be the covariance matrices of the source point cloud and the target point cloud, respectively.

[0104] Then, singular value decomposition is performed on the covariance matrix to obtain the eigenvectors of the source point cloud and the target point cloud, i.e., the principal axis directions. That is, for M... P With M Q After performing singular value decomposition, we obtain M. P With M Q The 3*3 feature vectors are the principal axis directions of the source point cloud P and the target point cloud Q, respectively.

[0105] Step S103b: Calculate the rotation matrix between the two sets of point clouds based on the principal axis directions of the two sets of point clouds.

[0106] Specifically, the rotation matrix is ​​calculated according to the following formula (3):

[0107] R = E P E Q -1 (3)

[0108] Step S103c: Update the centroid position of the source point cloud using the rotation matrix, and obtain the translation vector T between the two sets of point clouds based on the centroid offset of the two sets of point clouds.

[0109] In this embodiment, the centroid position of the source point cloud P is first updated using the rotation matrix R. That is, get Then, the translation vector T of the two sets of point clouds is calculated according to the following formula (4):

[0110]

[0111] Step S103d: Based on the rotation matrix and translation vector, an affine transformation matrix between the two sets of point clouds is obtained. Specifically, the affine transformation matrix is ​​obtained by multiplying the rotation matrix and translation vector, and then this affine transformation matrix is ​​used to map the location information of abnormal parts on the CBCT virtual model onto the three-dimensional digital model of the teeth.

[0112] As can be seen, this embodiment determines the centroid and principal axis direction of two sets of point clouds, and calculates the rotation matrix and translation vector between the two sets of point clouds accordingly. Based on the rotation matrix and translation vector between the centroids, the affine transformation matrix between the two sets of point clouds is obtained, thereby mapping the location information of abnormal parts on the CBCT virtual model to the three-dimensional digital model of the teeth. The registration scheme of this embodiment is simple, has low computational load, and can quickly map the location information of abnormal parts on the CBCT virtual model to the three-dimensional digital model of the teeth.

[0113] Generally, medications used to treat oral diseases are gradually released into the area corresponding to the drug storage structure and its surrounding area. Therefore, the location information of abnormal sites does not necessarily need to be very accurate. The location information of abnormal sites mapped onto the three-dimensional digital model of the teeth through the above registration is sufficient to meet the needs of subsequent treatments.

[0114] Of course, the higher the registration accuracy, the more accurate the information of abnormal parts obtained, and the better the effect of drug treatment. Therefore, in order to improve the registration accuracy, the registration method used in this embodiment may also include, but is not limited to, the congruent four-point set (4PCS) algorithm and its improved algorithm, the registration method based on local feature description, and the NDT registration method based on probability distribution. Since model registration is a mature technology in this field, it will not be elaborated here.

[0115] Of course, to further improve the registration accuracy and make the abnormal area information obtained on the three-dimensional digital model of the teeth more accurate, a secondary registration can be performed based on the above registration results. The secondary registration can select some reference points for the two models and use the Iterative Closest Point (ICP) algorithm to further register in order to improve the model registration results. Since the ICP algorithm has been widely used in registration, it will not be described in detail here.

[0116] The second embodiment of the present invention provides a method for generating a dental orthodontic treatment plan, the process of which is as follows: Figure 8 As shown, the method for generating a dental orthodontic treatment plan includes the following steps:

[0117] Step S801: Obtain a three-dimensional digital model of the teeth with abnormal location information of the patient's current dentition as a first digital tooth model representing the layout of the patient's first dentition. The abnormal location information includes at least the abnormal type and abnormal location information of the current patient.

[0118] The three-dimensional digital model of the tooth with abnormal location information can be obtained by the method for constructing a three-dimensional digital model of the tooth according to the first embodiment of the present invention. Since this process has been described in detail in the first embodiment, it will not be repeated in this embodiment.

[0119] Step S802: Determine the treatment-correction composite stage of the current patient's corrective plan based on the abnormality type;

[0120] Step S803: Obtain a second digital tooth model representing the current patient's second dentition layout;

[0121] Step S804: Based on the determined treatment and orthodontic complex stage and the first digital tooth model and the second digital tooth model, a series of target digital tooth models are obtained. The series of target digital tooth models represent a series of orthodontic states in which the patient's teeth are progressively moved from the first dentition layout to the second dentition layout. The series of target digital tooth models correspond to the treatment and orthodontic complex stage and the subsequent individual orthodontic stages. The target digital tooth model corresponding to the treatment and orthodontic complex stage has corresponding abnormality information.

[0122] Step S805: Based on the abnormal location information, generate corresponding digital models of drug storage structures on the target digital tooth models in the treatment and orthodontic complex stage.

[0123] As can be seen, this embodiment includes a treatment-orthodontic composite stage and a subsequent individual orthodontic stage for patients with abnormal sites. Based on the type of abnormality, the target digital tooth model corresponding to the treatment-orthodontic composite stage is determined, so that each target digital tooth model in the treatment-orthodontic composite stage has the corresponding abnormal site information. A corresponding drug storage structure digital model is generated for the abnormal site information, so that the shell-shaped orthodontic appliance prepared accordingly has the corresponding drug storage structure. This allows the medication to be stored in the treatment-orthodontic composite stage, enabling patients to receive treatment while undergoing orthodontic treatment, thereby improving the efficiency of the current patient's orthodontic treatment.

[0124] The following is a detailed description of steps S802-S805 of the second embodiment of the present invention:

[0125] When a patient's 3D digital model of their teeth contains information about abnormal areas, it indicates that the patient has an oral disease. In this embodiment, for such patients, the orthodontic treatment plan includes two phases: a treatment-orthodontic composite phase and a single-dental orthodontic phase. The treatment-orthodontic composite phase refers to treating the abnormal areas while simultaneously performing orthodontic treatment, while the single-dental orthodontic phase refers to treating only the teeth after the treatment of the abnormal areas has been completed. Treatment of abnormal areas requires a certain period of time, and the required treatment period often varies depending on the type of abnormality. The treatment period for the abnormal areas determines the treatment-orthodontic composite phase in the orthodontic treatment plan, that is, it determines the number of orthodontic steps corresponding to the treatment-orthodontic composite phase. Therefore, in step S802, it is necessary to determine the number of treatment steps in the treatment-orthodontic composite stage of the current patient's orthodontic plan based on the type of abnormality. In some examples, the doctor or designer can determine the treatment period required for the treatment of the abnormal site based on the type of abnormality, and then determine the number of treatment steps in the treatment-orthodontic composite stage. For example, when the current patient's abnormality type is periodontitis, the doctor can determine the treatment period required for the current patient's periodontitis treatment based on the severity of periodontitis. For example, if the treatment period is 2 months (60) days, and assuming that the time for each treatment step in the determined orthodontic plan is 7 days, then the treatment-orthodontic composite stage of the current patient's orthodontic plan is determined accordingly. In some examples, the treatment-orthodontic complex stage of the current patient's orthodontic plan can be determined based on the treatment plans of patients with the same abnormality type in historical data. For example, if the current patient's abnormality type is periodontitis, the treatment plans of periodontitis patients in historical data can be obtained, and the number of treatment steps corresponding to the treatment-orthodontic complex stage in their treatment plans can be used as the number of treatment steps in the treatment-orthodontic complex stage of the current patient's orthodontic plan. If there are multiple periodontitis patients, the average number of treatment steps in the treatment-orthodontic complex stage of the multiple patients' orthodontic plans can be used as the number of treatment steps in the treatment-orthodontic complex stage of the current patient's orthodontic plan.

[0126] In step S803, the second dental arch layout is the dental treatment target designed in the current orthodontic treatment plan. It can be the final target layout of the patient's dental arch treatment or a staged treatment target. The second digital dental model refers to the digital representation of the three-dimensional structure of the tooth state of the second dental arch layout.

[0127] Taking the final target layout as an example, in some implementations, the layout of the second row of teeth can be manually confirmed by clinicians or designers. For example, clinicians or designers determine the target layout of the patient's teeth according to the medical goals of the orthodontic treatment. This method of determining the target layout after treatment is simple and convenient, but it depends on the experience of clinicians or designers. Due to the different levels of clinicians or designers, the degree of control over the teeth arrangement results may vary.

[0128] In other embodiments, the second dental layout can also be completed semi-automatically or fully automatically based on a three-dimensional digital model of teeth representing the initial dental layout, according to medical rules and algorithms. For example, in some examples, the final target layout can be obtained based on arch curve fitting. That is, firstly, feature points on the teeth are used to fit an ideal arch curve, and then, according to the current position of each tooth in the patient's dentition and the positional relationship of the ideal arch curve, the teeth on the patient's digital dental model are arranged on the ideal arch curve to obtain the post-treatment target layout. In other examples, the post-treatment target layout can also be obtained based on deep learning methods. For example, a neural network model is first constructed, and a large number of digital dental models of the patient's dentition that conform to medical rules are used as training samples to train the neural network model. The trained neural network model can then obtain the target layout for the patient's dentition treatment based on a digital dataset representing the initial dental layout.

[0129] In step S804, after obtaining the first digital tooth model representing the current patient's first dentition layout and the second digital tooth model representing the current patient's second dentition layout, a series of digital datasets representing multiple intermediate tooth states as the patient moves from the first dentition layout to the second dentition layout can be obtained based on the first digital tooth model and the second digital tooth model. These datasets represent a series of target digital tooth models representing intermediate tooth states.

[0130] In some examples, the aforementioned successive intermediate state digital datasets can be obtained by determining the positional differences of each tooth between the first and second digital tooth models and interpolating these positional differences for each tooth. During orthodontic treatment, this interpolation is performed in multiple treatment steps, typically at least 10, sometimes 25, or even 40 or more. These positional differences correspond to the desired tooth movement. The interpolation can be linear for some or all of these positional differences, or it can be non-linear, for example, using a spline curve to determine the interpolation function in a conventional manner to achieve the interpolation between the positional differences. Since obtaining a series of target digital tooth models representing intermediate tooth states through interpolation is a standard technique in this field, the specific method for obtaining this series of target digital tooth models representing intermediate tooth states will not be elaborated upon here.

[0131] In the process of interpolating to obtain the series of target digital tooth models representing the intermediate tooth state, it is necessary to determine the series of target digital tooth models corresponding to the treatment and orthodontic composite stage obtained in step S803, as well as the series of target digital tooth models corresponding to the subsequent individual tooth orthodontic stages. For example, in step S803, it is determined that the treatment and orthodontic composite stage corresponds to 9 orthodontic steps. Since oral diseases need to be treated in a timely manner, the first 9 orthodontic steps in the series of target digital tooth models correspond to the treatment and orthodontic composite stage, while the tenth orthodontic step and subsequent orthodontic steps in the series of target digital tooth models correspond to the individual tooth orthodontic stages. The target digital tooth models corresponding to the treatment and orthodontic composite stage have corresponding abnormal site information, while the target digital tooth models corresponding to the individual tooth orthodontic stages can have the corresponding abnormal site information removed.

[0132] During orthodontic treatment, each tooth may move in different ways, including overall movement, rotation, torque, axial movement, elongation, and intrusion. For each movement, each orthodontic step corresponds to a specific single-step movement design. Generally, for the same movement, without considering differences in the diaphragm, the larger the single-step movement design, the greater the orthodontic force applied to the tooth by the appliance. Because patients experience discomfort in the oral cavity due to abnormalities during the treatment-orthodontic complex phase, excessively large single-step movement designs for any movement can further increase discomfort and even worsen the patient's condition. Therefore, in some implementations, the single-step movement design for each movement mode of the target digital tooth model in the treatment-orthodontic complex phase should be smaller than the single-step movement design for the corresponding movement mode of the target digital tooth model in the individual tooth orthodontic phase. Specifically, the single-step movement design for the movement mode of the target digital tooth model in the individual tooth orthodontic phase... Momentum refers to the designed single-step movement amount. For example, for a single-tooth orthodontic stage, the designed single-step movement amount for distal canine movement is 0.02mm. For a treatment-orthodontic complex stage, the designed single-step movement amount for distal canine movement is designed to be less than 0.02mm, such as 0.01mm. In this embodiment, the designed single-step movement amount for each movement mode of each target digital tooth model in the treatment-orthodontic complex stage is smaller than that in the single-tooth orthodontic stage. This allows the shell-shaped orthodontic appliance corresponding to the treatment-orthodontic complex stage to apply a smaller orthodontic force to the teeth, thereby avoiding strong discomfort for the patient when treatment and orthodontic treatment are carried out simultaneously.

[0133] In other implementations, as the treatment and correction are carried out simultaneously, the patient's affected area gradually heals due to the effects of the medication, and the discomfort caused by the affected area gradually decreases. Consequently, the corrective force that the patient can tolerate gradually increases. In this context, to improve the efficiency of orthodontic treatment, the single-step design movement amount of each movement mode in the target digital tooth model during the complex treatment phase can be increased progressively with the progress of the treatment plan. This increase can be phased; for example, in a complex treatment phase comprising nine steps, for canine distalization, the single-step design movement amount could be 0.01mm for the first three steps, 0.015mm for steps 4-6, and 0.018mm for steps 7-9. Alternatively, the increase can be gradual; similarly, for canine distalization, the first step's single-step design movement amount could be 0.01mm, the second 0.011mm, and the third 0.012mm, with each step gradually increasing. This incremental single-step design amount takes into account the actual situation of the patient's oral treatment, improving both treatment efficiency and patient comfort.

[0134] In step S805, for each target digital tooth model corresponding to the treatment and orthodontic complex stage, a corresponding drug storage structure digital model can be generated from the abnormal parts in the corresponding abnormal part information.

[0135] In this embodiment, each target digital tooth model in the treatment-orthodontic complex stage contains information on abnormal sites. To obtain a shell-shaped orthodontic appliance with a drug-filled structure, a digital model of the drug-filled structure needs to be generated for the abnormal site. In some implementations, to ensure that the drug-filled structure does not interfere with the orthodontic treatment, the digital model of the drug-filled structure can be a protrusion extending away from the tooth at the location of the abnormal site. The height of the protrusion determines the height of the corresponding drug-filled structure. This height will cause a foreign body sensation in the patient's mouth; therefore, the height of the protrusion is preferably limited to 0.5mm to 5mm. For drug treatment, if the effective area of ​​the drug treatment is too small, it will affect the treatment effect. Therefore, the protrusion covers at least 1 / 2 of the area of ​​the corresponding abnormal site to ensure the effective area of ​​the drug treatment and improve the treatment effect. At the same time, to further reduce the foreign body sensation when wearing the drug-filled structure formed by the protrusion, the edge of the drug-filled structure that contacts the oral cavity should be rounded to ensure a smooth transition at the contact point between the drug-filled structure and the oral cavity, reducing the foreign body sensation.

[0136] The third embodiment of the present invention provides a method for preparing a shell-shaped orthodontic appliance, the process of which is as follows: Figure 9 As shown, the method for preparing the shell-shaped orthodontic appliance includes the following steps:

[0137] Step S901: Obtain a series of target digital tooth models representing the gradual movement of the patient's dentition from the first dentition layout to the second dentition layout. The series of target digital tooth models correspond to the treatment-orthodontic complex stage and the subsequent individual orthodontic stages. Each target digital tooth model in the treatment-orthodontic complex stage has abnormal site information and a digital model of the drug storage structure corresponding to the abnormal site information.

[0138] The method for generating an orthodontic treatment plan according to a second embodiment of the present invention obtains a series of target digital tooth models representing the gradual movement of the patient's dentition from the first dentition layout to the second dentition layout. This process has been described in detail in the second embodiment and will not be repeated in this embodiment.

[0139] Step S902: Based on each of the target digital tooth models, a series of shell-shaped orthodontic appliances are prepared, wherein each shell-shaped orthodontic appliance corresponding to the treatment and orthodontic composite stage has a drug storage structure on its shell corresponding to the abnormal part for storing drugs.

[0140] Step S903: Based on the current patient's abnormality type, add corresponding medication to the medication storage structure of each shell-shaped orthodontic appliance corresponding to the treatment-orthodontic complex stage, so as to provide medication to the abnormal site when worn on the patient's dentition during the treatment-orthodontic complex stage.

[0141] As can be seen, the digital model of the drug storage structure on the digital tooth model of each target in the treatment-orthodontic complex stage can be directly used to prepare a shell-shaped orthodontic appliance with a drug storage structure. By adding drugs targeting the abnormal type inside the drug storage structure, the abnormal area can be treated at the same time as the teeth are being corrected when the patient wears the shell-shaped orthodontic appliance in the treatment-orthodontic complex stage. The correction and treatment are carried out simultaneously, thereby improving the efficiency of the patient's orthodontic treatment.

[0142] The following describes in detail steps S902-S903 of the third embodiment of the present invention:

[0143] For step S902, in some examples, the hot-press molding process can be used to manufacture shell-shaped dental appliances. The process is as follows: 3D printing is performed on the series of target digital tooth models to create a solid dental model. The solid dental model corresponding to the treatment-orthodontic complex stage has a physical drug storage structure model corresponding to the abnormal area. Then, hot-press molding is performed on the solid dental model to obtain a shell-shaped dental instrument containing the shape of teeth. The shell of each shell-shaped dental appliance in the treatment-orthodontic complex stage forms a drug storage structure for storing drugs on the corresponding physical drug storage structure model.

[0144] In other examples, when using additive manufacturing to prepare shell-shaped orthodontic appliances, the specific manufacturing process involves designing a series of digital shell-shaped orthodontic appliance models based on a series of target digital tooth models. Corresponding to the target digital tooth model in the treatment-orthodontic complex stage, a corresponding shell-shaped orthodontic appliance model is also formed based on its drug storage structure digital model. Then, the designed digital shell-shaped orthodontic appliance models are printed using 3D printing. In the treatment-orthodontic complex stage, each shell-shaped orthodontic appliance has a drug storage structure formed on the shell corresponding to the abnormal area for storing medication.

[0145] Once the shell-shaped orthodontic appliance with a drug storage structure corresponding to the treatment and orthodontic complex stage is obtained, medication needs to be added to it to achieve the therapeutic effect. Therefore, in step S903, a targeted treatment medication needs to be selected according to the current patient's abnormality type and added to the drug storage structure of the shell-shaped orthodontic appliance in the treatment and orthodontic complex stage. This allows the abnormal area to receive medication treatment while the orthodontic appliance is worn in the treatment and orthodontic complex stage.

[0146] Generally, the greater the elasticity of the diaphragm used to prepare the shell-shaped orthodontic appliance, the greater the elasticity of the resulting shell-shaped appliance. A more elastic shell-shaped appliance has a relatively smaller overall corrective force, resulting in better patient comfort. Since the treatment-treatment complex stage can cause discomfort due to abnormal areas on the patient's body, the elasticity of the shell-shaped appliances in the treatment-treatment complex stage should be greater than that of the shell-shaped appliances in the individual orthodontic stages to minimize patient discomfort and improve overall comfort.

[0147] As the discomfort in the abnormal area gradually decreases with treatment, in order to improve the efficiency of orthodontic treatment, the elasticity of the shell of each shell-shaped orthodontic appliance in the treatment complex stage can be selected to gradually decrease with the treatment sequence, thereby gradually increasing the orthodontic force of the shell-shaped orthodontic appliance, thus ensuring the achievement of the orthodontic effect and improving the efficiency of orthodontic treatment.

[0148] The following example uses a hot-press molding process to illustrate the shell-shaped orthodontic appliance with a drug storage structure obtained in this embodiment:

[0149] In some examples, due to the existence of a physical drug storage structure model, after the thermoforming process, the receiving cavity of the shell-shaped orthodontic appliance 10 partially protrudes outward to form a drug storage structure 110 corresponding to the physical drug storage structure model. That is, the drug storage structure 110 is integrally formed with the shell-shaped orthodontic appliance 10. Then, the medication 111, such as a pill or tablet, is fixed inside the drug storage structure 110. The side of the drug storage structure 110 facing the tooth 11 can be directly open. Figure 10 As shown, drug 111 can be directly and slowly released to the corresponding abnormal site through the opening. Direct release of drug 111 through the opening allows for a larger dose, increasing the treatment speed for oral diseases. Of course, after placing drug 111 within the drug storage structure 110, a sustained-release layer 112 can also be provided on the side of the drug storage structure 111 facing the tooth 11, such as... Figure 11 As shown, the sustained-release layer 112 can confine the drug 111 within the storage compartment, and can also slowly release the drug 111 to the patient's abnormal site through pores or a mesh structure. Furthermore, releasing the drug through the sustained-release layer 112 can reduce the drug release rate, increase the duration of action of the drug in the oral cavity, and improve the treatment effect.

[0150] For some oral diseases, the abnormal location does not correspond to the tooth position, but rather to the patient's gum position. For such abnormal locations, the physical drug storage structure is also located at the corresponding gum position. In this case, the receiving cavity of the shell-shaped orthodontic appliance after thermoforming will include a tooth receiving cavity 120 and a gum receiving cavity 130, such as... Figure 12 As shown, the tooth receiving cavity 120 is used to receive the patient's clinical crown, and the gingival receiving cavity 130 is used to receive the gingival soft tissue. After the thermo-pressed film is formed, the gingival receiving cavity of the shell-shaped orthodontic appliance corresponds to the physical drug storage structure model, which locally protrudes to form the drug storage structure 110. Then, the drug 111 is fixed in the drug storage structure 110.

[0151] The fourth embodiment of the present invention also provides an electronic device, including at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the method for constructing a three-dimensional digital model of teeth as described above or a method for generating an orthodontic treatment plan.

[0152] The memory and processor are connected via a bus, which can include any number of interconnecting buses and bridges, connecting various circuits of one or more processors and memories. The bus can also connect various other circuits, such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and will not be described further herein. The bus interface provides an interface between the bus and the transceiver. The transceiver can be a single element or multiple elements, such as multiple receivers and transmitters, providing a unit for communicating with various other devices over a transmission medium. Data processed by the processor is transmitted over the wireless medium via an antenna, which further receives data and transmits it to the processor.

[0153] The processor manages the bus and general processing, and also provides various functions, including timing, peripheral interfaces, voltage regulation, power management, and other control functions. Memory is used to store data used by the processor during operation.

[0154] Another embodiment of the present invention relates to a computer-readable storage medium storing a computer program. When executed by a processor, the computer program implements the above-described method for constructing a three-dimensional digital model of teeth or a method for generating an orthodontic treatment plan.

[0155] That is, those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. This program is stored in a storage medium and includes several instructions to cause a device (which may be a microcontroller, chip, etc.) or processor to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0156] Those skilled in the art will understand that the above embodiments are specific embodiments for implementing this application, and in practical applications, various changes can be made to them in form and detail without departing from the spirit and scope of this application.

Claims

1. A method for constructing a three-dimensional digital model of teeth, characterized in that, Includes the following steps: Acquire CBCT imaging data of the patient's current oral cavity and a three-dimensional digital model of the teeth representing the patient's current dentition; The CBCT image data is segmented using a pre-trained segmentation model of abnormal parts of teeth and oral cavity to obtain a CBCT virtual model of the patient's current dentition and the abnormal part identification results of the CBCT virtual model. The abnormal part identification results include at least the abnormality type and abnormality location information. The CBCT virtual model with the identified abnormal parts is registered with the three-dimensional digital model of the tooth. Based on the registration result, the abnormal part identification result in the CBCT virtual model is mapped to the three-dimensional digital model of the tooth to obtain a three-dimensional digital model of the tooth with abnormal part information. The abnormal part information includes at least the abnormal type and abnormal location information.

2. The method for constructing a three-dimensional digital model of teeth as described in claim 1, characterized in that, The segmentation model for teeth and oral cavity abnormalities is trained from multiple pre-acquired training samples, and the training includes: Based on historical data from multiple patients, several sample CBCT images were selected for different categories of dental abnormalities and labeled with features to construct a sample CBCT image dataset. The label feature annotation includes target localization annotation and abnormality classification annotation. Image features of training samples in the CBCT image dataset are extracted using a feature extraction model. The image features are identified using a feature recognition model to obtain the target recognition result of the training sample. The target result includes the tooth classification result and the abnormal part recognition result.

3. The method for constructing a three-dimensional digital model of teeth as described in claim 2, characterized in that, The segmentation model for teeth and oral cavity abnormalities is a neural network model for two-dimensional image segmentation. The training sample is input into the segmentation model for teeth and oral cavity abnormalities as a two-dimensional slice sequence to obtain the recognition result of each two-dimensional slice. The recognition results of all two-dimensional slices are stacked to obtain the target recognition result of the training sample.

4. The method for constructing a three-dimensional digital model of teeth as described in claim 2, characterized in that, The segmentation model for teeth and oral cavity abnormalities is a neural network model for three-dimensional image segmentation. The training samples are obtained by inputting three-dimensional images into the segmentation model for teeth and oral cavity abnormalities.

5. The method for constructing a three-dimensional digital model of teeth as described in any one of claims 1-4, characterized in that, The abnormality type includes at least one of periodontitis, pulpitis, gingivitis, dental caries, alveolar osteitis, and periapical periodontitis.

6. The method for constructing a three-dimensional digital model of teeth as described in claim 1, characterized in that, The registration of the CBCT virtual model with identified abnormal areas to the three-dimensional digital model of the teeth, and the mapping of the abnormal area identification results in the CBCT virtual model to the three-dimensional digital model of the teeth based on the registration results, includes: The point cloud data of the CBCT virtual model and the three-dimensional digital model of the teeth are obtained as the source point cloud and the target point cloud, and the principal axis directions of the two sets of point clouds are extracted. Calculate the rotation matrix between the two sets of point clouds based on the principal axis directions of the two sets of point clouds; Based on the rotation matrix and translation vector, the affine transformation matrix between the two sets of point clouds is obtained; The affine transformation matrix is ​​used to map the abnormal area information of the CBCT virtual model to the three-dimensional digital model of the teeth.

7. A method for generating orthodontic treatment plans, characterized in that, Includes the following steps, The method for constructing a three-dimensional digital model of teeth according to any one of claims 1-6 obtains a three-dimensional digital model of teeth of the current patient with information on abnormal tooth locations as a first digital tooth model representing the layout of the first dentition of the patient. The information on abnormal tooth locations includes at least the abnormal type and abnormal location information of the current patient. The treatment and correction phases of the current patient's treatment plan are determined based on the type of abnormality. Obtain a second digital tooth model representing the current patient's second dentition layout; Based on the determined treatment and orthodontic complex stage and the first and second digital tooth models, a series of target digital tooth models are obtained. The series of target digital tooth models represent a series of orthodontic states in which the patient's teeth are progressively moved from the first dentition layout to the second dentition layout. The series of target digital tooth models correspond to the treatment and orthodontic complex stage and the subsequent individual orthodontic stages. Each target digital tooth model corresponding to the treatment and orthodontic complex stage has corresponding abnormality information. Based on the information about the abnormal locations, corresponding digital models of drug storage structures are generated on the target digital tooth models in the combined treatment and orthodontic stage.

8. The method for generating an orthodontic treatment plan as described in claim 7, characterized in that, In the combined treatment and orthodontic phase, the single-step design movement amount of each movement mode of each target digital tooth model is less than the single-step design movement amount of the corresponding movement mode of each target digital tooth model in the individual orthodontic phase.

9. The method for generating an orthodontic treatment plan as described in claim 8, characterized in that, In the treatment and orthodontic complex phase, the single-step design of the movement amount of each movement mode of the target digital tooth model increases as the treatment plan progresses.

10. The method for generating an orthodontic treatment plan as described in claim 7, characterized in that, The digital model of the drug storage structure covers at least half of the area of ​​the corresponding abnormal part.

11. A method for preparing a shell-shaped orthodontic appliance, characterized in that, The preparation method includes: The method for generating a dental treatment plan based on any one of claims 7-10 obtains a series of target digital tooth models representing the gradual movement of a patient's dentition from a first dentition layout to a second dentition layout, the series of target digital tooth models corresponding to the treatment-orthodontic complex phase and subsequent individual dental treatment phases, each target digital tooth model in the treatment-orthodontic complex phase having abnormal site information and a digital model of a drug storage structure corresponding to the abnormal site information. A series of shell-shaped orthodontic appliances are prepared based on the target digital tooth models, wherein each shell-shaped orthodontic appliance corresponding to the treatment and orthodontic composite stage has a drug storage structure on its shell corresponding to the abnormal site for storing drugs. Based on the current patient's abnormality type, corresponding medications are added to the medication storage structures of each shell-shaped orthodontic appliance corresponding to the treatment-orthodontic complex stage, so as to provide medication to the abnormal site when the appliance is worn on the patient's dentition during the treatment-orthodontic complex stage.

12. The method for preparing the shell-shaped orthodontic appliance as described in claim 11, characterized in that, The elasticity of the shell of each shell-shaped orthodontic appliance in the treatment-orthodontic complex stage is greater than that of the shell of each shell-shaped orthodontic appliance in the individual orthodontic stage.

13. The method for preparing the shell-shaped orthodontic appliance as described in claim 12, characterized in that, The elasticity of the shell of each shell-shaped orthodontic appliance in the treatment and orthodontic complex stage gradually decreases as the treatment sequence progresses.

14. An electronic device, characterized in that, include: At least one processor; as well as, A memory communicatively connected to the at least one processor; wherein, The memory stores instructions executable by the at least one processor, which enable the at least one processor to perform a method for constructing a three-dimensional digital model of teeth as claimed in any one of claims 1 to 6 or a method for generating an orthodontic treatment plan as claimed in any one of claims 7 to 10.

Images