A method and system for predicting development of permanent teeth

Abstract

The application discloses a method and system for predicting permanent tooth development, and relates to the field of tooth development prediction. The method comprises the following steps: acquiring an oral CBCT to be staged; performing segmentation processing on the oral CBCT to be staged to obtain oral CBCTs of different teeth; encoding the oral CBCTs of different teeth by using a tooth position representation method to obtain the oral CBCTs of different teeth after encoding; determining permanent anterior teeth, permanent premolars and permanent molars to be staged according to the oral CBCTs of different teeth after encoding and the root morphologies of different teeth; and inputting the permanent anterior teeth, the permanent premolars and the permanent molars to be staged into corresponding prediction models respectively to obtain corresponding calcification stages. The application can improve the prediction accuracy by classifying the teeth and inputting the classified teeth into corresponding prediction models.

Description

Technical Field

[0001] This invention relates to the field of predicting tooth development, and in particular to a method and system for predicting permanent tooth development. Background Technology

[0002] The mixed dentition stage is the transitional period from deciduous to permanent dentition, accompanied by jawbone growth, development of the permanent tooth germ, physiological resorption of deciduous tooth roots, remodeling of surrounding alveolar bone, and growth and functional establishment of soft tissues. Tooth development includes the growth phase, calcification phase, and eruption phase. Nolla, through radiographic changes, divided the permanent tooth calcification process into ten stages, serving as a commonly used clinical reference indicator for assessing the degree of tooth development. For the same disease, dentists will choose different treatment methods at different stages of permanent tooth development. For example, for periapical lesions of permanent teeth, if the permanent tooth development has not reached stage 8, pulp revascularization is the first choice; at stage 9, apexification or periapical barrier surgery is chosen; and at stage 10, root canal treatment is chosen. Therefore, accurate assessment of tooth development stages is of great guiding significance for clinical decision-making.

[0003] In existing technologies, a single neural network is used to predict the developmental stages of multiple types of teeth. However, each type of tooth differs at different developmental stages. Therefore, using a single neural network to predict the developmental stages of multiple types of teeth can lead to inaccurate prediction results for each type of tooth development stage. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for predicting permanent tooth development. By classifying teeth and inputting the classified teeth into the corresponding prediction model, the accuracy of the prediction can be improved.

[0005] To achieve the above objectives, the present invention provides the following solution:

[0006] A method for predicting permanent tooth development, the method comprising:

[0007] Obtain the oral CBCT scan for staging;

[0008] The oral CBCT of the stage to be segmented is processed to obtain oral CBCT of different teeth;

[0009] The oral CBCT of different teeth was encoded using tooth position representation to obtain oral CBCT of different teeth;

[0010] Based on the oral CBCT scans of different teeth and the root morphology of different teeth, the permanent anterior teeth to be staged, the permanent premolars to be staged, and the permanent molars to be staged were determined.

[0011] The permanent anterior teeth to be staged are input into the permanent anterior tooth prediction model to obtain the calcification stage of the permanent anterior teeth.

[0012] The permanent premolars to be staged are input into the permanent premolar prediction model to obtain the calcification stage of the permanent premolars.

[0013] The permanent molars to be staged are input into the permanent molar prediction model to obtain the calcification stage of the permanent molars.

[0014] Optionally, the oral CBCT to be staged is segmented to obtain oral CBCTs of different teeth, specifically including:

[0015] The open-source annotation tool 3D Slicer was used to segment the oral CBCT of the teeth to be staged, resulting in oral CBCT of different teeth.

[0016] Optionally, the permanent anterior teeth to be staged are input into the permanent anterior tooth prediction model to obtain the calcification stage of the permanent anterior teeth, which also includes:

[0017] Construct a convolutional neural network;

[0018] Acquire CBCT training data for permanent anterior teeth; the CBCT training data for permanent anterior teeth includes CBCT samples of permanent anterior teeth and the corresponding calcification stages of the CBCT samples of permanent anterior teeth.

[0019] The permanent anterior dentistry CBCT training data was divided into a training set and a validation set.

[0020] The weights and biases of the convolutional neural network are trained using the training set with the aim of minimizing the loss function, resulting in a trained convolutional neural network; the loss function is determined based on the error between the output of the convolutional neural network and the calcification stage corresponding to the CBCT sample of the permanent anterior teeth.

[0021] The weights and biases of the trained convolutional neural network are adjusted using the validation set to obtain the adjusted convolutional neural network.

[0022] The adjusted convolutional neural network was used as the prediction model for the permanent anterior teeth.

[0023] Optionally, the tooth position representation method includes: FDI tooth position representation method, location recording method, and general recording method.

[0024] A system for predicting permanent tooth development, said system being applied to the method for predicting permanent tooth development described above, said system comprising:

[0025] The acquisition module is used to acquire the oral CBCT scan to be staged.

[0026] The segmentation module is used to segment the oral CBCT to be staged, resulting in oral CBCTs of different teeth.

[0027] The encoding module is used to encode the oral CBCT of different teeth using tooth position representation, so as to obtain the oral CBCT after encoding different teeth;

[0028] The classification module is used to determine the permanent anterior teeth, permanent premolars, and permanent molars to be staged based on the oral CBCT scans of different teeth and the root morphology of different teeth.

[0029] The permanent anterior tooth prediction module is used to input the permanent anterior teeth to be staged into the permanent anterior tooth prediction model to obtain the calcification stage of the permanent anterior teeth.

[0030] The permanent premolar prediction module is used to input the permanent premolar to be staged into the permanent premolar prediction model to obtain the calcification stage of the permanent premolar.

[0031] The permanent molar prediction module is used to input the permanent molar to be staged into the permanent molar prediction model to obtain the calcification stage of the permanent molar.

[0032] Optionally, the system for predicting permanent tooth development further includes: a training module, specifically comprising:

[0033] Building units are used to construct convolutional neural networks;

[0034] The acquisition unit is used to acquire CBCT training data of the permanent anterior teeth oral cavity; the CBCT training data of the permanent anterior teeth oral cavity includes CBCT samples of the permanent anterior teeth oral cavity and the calcification stage corresponding to the CBCT samples of the permanent anterior teeth oral cavity.

[0035] A classification unit is used to divide the permanent anterior dentist oral CBCT training data into a training set and a validation set;

[0036] The training unit is used to train the weights and biases of the convolutional neural network using the training set with the aim of minimizing the loss function, so as to obtain the trained convolutional neural network; the loss function is determined based on the error between the output of the convolutional neural network and the calcification stage corresponding to the CBCT sample of the permanent anterior teeth.

[0037] The verification unit is used to adjust the weights and biases of the trained convolutional neural network using the verification set to obtain the adjusted convolutional neural network.

[0038] The output unit is used to use the adjusted convolutional neural network as the prediction model for the permanent anterior teeth.

[0039] An electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method for predicting permanent tooth development as described above.

[0040] A computer-readable storage medium storing a computer program that, when executed, implements the method for predicting permanent tooth development as described above.

[0041] According to specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0042] This invention discloses a method and system for predicting permanent tooth development. The method includes: acquiring a CBCT (Cellular Brain-Cellular CT) of the teeth to be staged; segmenting the CBCT of the teeth to be staged to obtain CBCTs of different teeth; encoding the CBCTs of different teeth using a tooth position representation method to obtain coded CBCTs of different teeth; determining the permanent anterior teeth, permanent premolars, and permanent molars to be staged based on the coded CBCTs of different teeth and the root morphology of different teeth; inputting the permanent anterior teeth to be staged into a permanent anterior tooth prediction model to obtain the calcification stage of the permanent anterior teeth; inputting the permanent premolars to be staged into a permanent premolar prediction model to obtain the calcification stage of the permanent premolars; and inputting the permanent molars to be staged into a permanent molar prediction model to obtain the calcification stage of the permanent molars. This invention improves the accuracy of prediction by classifying teeth and inputting the classified teeth into the corresponding prediction models. Attached Figure Description

[0043] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0044] Figure 1 This is a flowchart of a method for predicting permanent tooth development in an embodiment of the present invention. Detailed Implementation

[0045] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0046] The purpose of this invention is to provide a method and system for predicting permanent tooth development. By classifying teeth and inputting the classified teeth into the corresponding prediction model, the accuracy of the prediction can be improved.

[0047] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0048] Example 1

[0049] like Figure 1 As shown, the present invention provides a method for predicting permanent tooth development, the method comprising:

[0050] Step 101: Obtain the oral CBCT scan to be staged.

[0051] Step 102: Segment the oral CBCT to be staged to obtain oral CBCTs of different teeth.

[0052] Step 103: Use tooth position representation to encode the oral CBCT of different teeth to obtain the oral CBCT after different tooth coding.

[0053] Step 104: Based on the oral CBCT scans of different teeth and the root morphology of different teeth, determine the permanent anterior teeth to be staged, the permanent premolars to be staged, and the permanent molars to be staged.

[0054] In practical application, dentists assign labels to the tooth segmentation results according to the FDI tooth position representation method and classify them according to root type into five categories: primary anterior teeth, primary molars, permanent anterior teeth, permanent premolars, and permanent molars. Based on the above category information, permanent tooth germs (permanent anterior teeth, permanent premolars, and permanent molars) are selected. According to the different root morphological characteristics presented by the anterior teeth, premolars, and molars at the stage diagnosis, they are divided into anterior teeth (single root), premolars (flat root), and molars (multiple roots).

[0055] Step 105: Input the permanent anterior teeth to be staged into the permanent anterior tooth prediction model to obtain the calcification stage of the permanent anterior teeth. Input the permanent premolars to be staged into the permanent premolar prediction model to obtain the calcification stage of the permanent premolars. Input the permanent molars to be staged into the permanent molar prediction model to obtain the calcification stage of the permanent molars.

[0056] As a specific embodiment, the oral CBCT to be staged is segmented to obtain oral CBCTs of different teeth, specifically including:

[0057] The open-source annotation tool 3D Slicer was used to segment the oral CBCT of the teeth to be staged, resulting in oral CBCT of different teeth.

[0058] In practical applications, the open-source annotation tool 3D Slicer is first used to perform interactive and rapid segmentation of CBCT images, using methods including seed segmentation and flood filling. Subsequently, the doctor performs fine delineation of the tooth region based on the coarse segmentation results to obtain the segmentation results of the teeth in the CBCT.

[0059] Specific segmentation methods can include seed segmentation or flood filling.

[0060] Seed segmentation, also known as region growing, is an image segmentation method based on pixel similarity. This method selects some seed points as starting points and then, according to certain criteria such as pixel color or pixel intensity, adds all pixels connected to the seed points that satisfy these conditions to the same region.

[0061] The core steps of seed segmentation are as follows:

[0062] 1) Select one or more seed points.

[0063] 2) Determine neighboring points similar to the seed point based on a certain similarity criterion (such as color, intensity, texture, etc.).

[0064] 3) Add these similar points to the region where the seed point is located, and use them as new seed points.

[0065] 4) Repeat steps 2 and 3 until a stopping condition is met (e.g., no more pixels can be added to the area), or the entire image has been checked.

[0066] Seed segmentation can be used in many different scenarios, such as image noise removal, image segmentation, and object detection. Common algorithm implementations include color-based region growing and texture-based region growing.

[0067] Flood fill is an algorithm frequently used in computer graphics. Its main purpose is to identify a connected region of a graphic or to replace a region with the same color or properties.

[0068] The name "Flood Fill" comes from how it works. Like a flood, it spreads outwards from a single point until certain conditions are met, such as matching colors or reaching a boundary. This is similar to using the fill tool in a drawing app to fill in color; if there are pixels of the same color around the area you click, they will be replaced with the new color.

[0069] The basic steps of the flood filling algorithm are as follows:

[0070] 1) Start from an initial point (seed point).

[0071] 2) Check the neighbors of the seed point (usually in four or eight directions).

[0072] 3) If neighboring points meet the criteria (usually the same color or within a certain threshold), change their color and add them to the list of points to be reviewed.

[0073] 4) Continue this process until there are no more points to check in the list.

[0074] Flood fill algorithms are used in many applications, such as computer graphics rendering, game map generation, medical image processing, and machine vision.

[0075] The prediction models for permanent anterior teeth, permanent premolars, and permanent molars are all obtained through training, and the training processes are similar, differing only in the inputs and outputs. Therefore, the training process of the permanent anterior tooth prediction model will be used as an example for introduction, as follows:

[0076] The method for obtaining the prediction model for permanent anterior teeth is as follows:

[0077] The permanent anterior teeth to be staged are input into the permanent anterior tooth prediction model to obtain the calcification stage of the permanent anterior teeth. This process also includes:

[0078] Construct a convolutional neural network.

[0079] Obtain CBCT training data for the permanent anterior teeth oral cavity; this data includes CBCT samples of the permanent anterior teeth oral cavity and the corresponding calcification stages. The calcification stages corresponding to the CBCT samples of the permanent anterior teeth oral cavity can be determined in various ways, such as:

[0080] 1) Age: The growth and development of children is evaluated by understanding their actual age and the state of their oral and overall growth and development.

[0081] 2) Bone age: Wrist bone X-rays are used clinically to evaluate the oral and overall developmental status of children.

[0082] 3) Dental age: The developmental status of teeth is used to evaluate a child's growth and development.

[0083] The CBCT training data for the permanent anterior teeth were divided into a training set and a validation set.

[0084] The weights and biases of the convolutional neural network are trained using the training set with the aim of minimizing the loss function, resulting in a trained convolutional neural network. The loss function is determined based on the error between the output of the convolutional neural network and the calcification stage corresponding to the CBCT sample of the permanent anterior teeth.

[0085] The weights and biases of the trained convolutional neural network are adjusted using the validation set to obtain the adjusted convolutional neural network.

[0086] The adjusted convolutional neural network was used as the prediction model for the permanent anterior teeth.

[0087] For specific applications and data preparation: First, due to the frequent presence of abnormally large and small isolated values ​​in CBCT images caused by metal artifacts, the foreground intensity values ​​were cropped to the percentage range of [0.5, 99.5]. Then, based on the mean and standard deviation of the foreground intensity values ​​across the entire dataset, z-socre was applied to normalize the intensity to the interval [0, 255]. In addition, several data augmentation strategies were used, including rotation, scaling, and mirroring.

[0088] Model Construction: The ResNet-50 neural network model was selected, and its network structure was constructed. It features 50 convolutional layers, including stacked residual blocks and global average pooling layers. Each residual block contains multiple convolutional layers and batch normalization layers, as well as cross-layer connections. This design allows the network to train deeper models more efficiently and reduces the number of parameters while maintaining accuracy. The main characteristics of ResNet-50 are its strong feature extraction capabilities and good model expressive power, enabling it to capture complex features in images. It is widely used in image recognition and has become the foundational model for many computer vision tasks.

[0089] Loss function definition: The L2 loss function, also known as Mean Squared Error (MSE), is a commonly used loss function, typically used in model training for regression problems. It measures the squared difference between the model's predicted values ​​and the true labels.

[0090] The L2 loss function is calculated as follows: L2 loss = (predicted value - actual value)^2.

[0091] Specifically, for each sample, a prediction is first obtained using the model's predicted value, and then this prediction is compared with the true label. The greater the difference, the greater the loss. The square of the difference is used to ensure that the loss value is always non-negative.

[0092] The advantage of the L2 loss function is that it imposes a larger penalty on samples with large prediction errors because the square amplifies the impact of the discrepancies, thus making the model pay more attention to samples with larger errors. Furthermore, the L2 loss function has favorable mathematical properties, such as differentiability and convexity, which facilitates model training using optimization algorithms such as gradient descent.

[0093] During training, the goal is to adjust the model's parameters by minimizing the L2 loss function, so that the difference between the model's predicted values ​​and the true labels is as small as possible. Through iterative training, the model continuously optimizes its parameters to reduce the L2 loss, thereby improving the model's predictive performance.

[0094] Optimizer selection: The Adam optimization algorithm is selected to update the model's weights and biases to minimize the loss function.

[0095] Batch training: The training data is divided into mini-batch data. The input data is forward-propagated on each batch, the loss function is calculated, the gradient is calculated through the backpropagation algorithm, and the model parameters are updated using the optimization algorithm.

[0096] Iterative training: The model is trained through multiple training iterations (epochs). In each iteration, the model parameters are updated using different batches of data until a predetermined number of training epochs are reached or a stopping criterion is met.

[0097] Model Evaluation: The trained model was evaluated using a validation set or a test set. The Mean-Square Error (MSE) metric was used to verify the accuracy of permanent tooth germ staging.

[0098] Hyperparameter tuning: Adjust the values ​​of hyperparameters, such as learning rate, regularization parameter, batch size, etc., based on the model evaluation results to optimize model performance.

[0099] Model saving and application: During training, the model's weights and parameters can be saved periodically for subsequent predictions and applications. The saved model can be loaded onto new data for prediction or used in practical applications.

[0100] The model was tested using a test set, and the accuracy of its permanent tooth developmental staging was evaluated using mean squared error (MSE).

[0101] The accuracy of permanent tooth germ staging was verified using the mean-square error (MSE) index.

[0102]

[0103] Where n represents the number of samples, i represents the i-th sample, f(x) is the predicted value, y is the true value, and MSE represents the deviation between the predicted value and the true value, reflecting the predictive ability of the model.

[0104] As a specific embodiment, the tooth position representation method includes: FDI tooth position representation method, location recording method, and general recording method.

[0105] Example 2

[0106] A system for predicting permanent tooth development, wherein the system is applied to the method for predicting permanent tooth development described in Example 1, the system comprising:

[0107] The acquisition module is used to acquire the oral CBCT scan to be staged.

[0108] The segmentation module is used to segment the oral CBCT to be staged, resulting in oral CBCTs of different teeth.

[0109] The encoding module is used to encode the oral CBCT of different teeth using tooth position representation, so as to obtain the oral CBCT after encoding different teeth.

[0110] The classification module is used to determine the permanent anterior teeth, permanent premolars, and permanent molars to be staged based on the oral CBCT scans of different teeth and the root morphology of different teeth.

[0111] The permanent anterior tooth prediction module is used to input the permanent anterior teeth to be staged into the permanent anterior tooth prediction model to obtain the calcification stage of the permanent anterior teeth.

[0112] The permanent premolar prediction module is used to input the permanent premolar to be staged into the permanent premolar prediction model to obtain the calcification stage of the permanent premolar.

[0113] The permanent molar prediction module is used to input the permanent molar to be staged into the permanent molar prediction model to obtain the calcification stage of the permanent molar.

[0114] As one specific embodiment, the system for predicting permanent tooth development further includes: a training module, specifically comprising:

[0115] Building units are used to construct convolutional neural networks.

[0116] The acquisition unit is used to acquire CBCT training data of the permanent anterior teeth oral cavity; the CBCT training data of the permanent anterior teeth oral cavity includes CBCT samples of the permanent anterior teeth oral cavity and the calcification stage corresponding to the CBCT samples of the permanent anterior teeth oral cavity.

[0117] The classification unit is used to divide the permanent anterior dental CBCT training data into a training set and a validation set.

[0118] The training unit is used to train the weights and biases of the convolutional neural network using the training set with the aim of minimizing the loss function, so as to obtain the trained convolutional neural network; the loss function is determined based on the error between the output of the convolutional neural network and the calcification stage corresponding to the CBCT sample of the permanent anterior teeth.

[0119] The verification unit is used to adjust the weights and biases of the trained convolutional neural network using the verification set to obtain the adjusted convolutional neural network.

[0120] The output unit is used to use the adjusted convolutional neural network as the prediction model for the permanent anterior teeth.

[0121] An electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the method for predicting permanent tooth development as described in Example 1.

[0122] A computer-readable storage medium storing a computer program that, when executed, implements the method for predicting permanent tooth development as described in Example 1.

[0123] The core invention of this application is that, based on the prior differences in tooth root morphology, three parallel neural network classification models are designed, with different models selected for different types of teeth as input. Without this design, using a single neural network model would result in a decrease in predictive performance due to the differences in scale and morphology among different types of teeth.

[0124] This invention improves the accuracy of predictions by classifying teeth and inputting the classified teeth into the corresponding prediction model.

[0125] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the systems disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the descriptions are relatively simple; relevant parts can be referred to the method section.

[0126] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for predicting permanent tooth development, characterized in that, The method for predicting permanent tooth development includes: Obtain the oral CBCT scan for staging; The oral CBCT of the stage to be segmented is processed to obtain oral CBCT of different teeth; The oral CBCT of different teeth was encoded using tooth position representation to obtain oral CBCT of different teeth; Based on the oral CBCT scans of different teeth and the root morphology of different teeth, the permanent anterior teeth to be staged, the permanent premolars to be staged, and the permanent molars to be staged were determined. The permanent anterior teeth to be staged are input into the permanent anterior tooth prediction model to obtain the calcification stage of the permanent anterior teeth. The permanent premolars to be staged are input into the permanent premolar prediction model to obtain the calcification stage of the permanent premolars. The permanent molars to be staged are input into the permanent molar prediction model to obtain the calcification stage of the permanent molars; The method for obtaining the permanent anterior tooth prediction model is as follows: Construct a convolutional neural network; Acquire CBCT training data for permanent anterior teeth; the CBCT training data for permanent anterior teeth includes CBCT samples of permanent anterior teeth and the corresponding calcification stages of the CBCT samples of permanent anterior teeth. The permanent anterior dentistry CBCT training data was divided into a training set and a validation set. The weights and biases of the convolutional neural network are trained using the training set with the aim of minimizing the loss function, resulting in a trained convolutional neural network; the loss function is determined based on the error between the output of the convolutional neural network and the calcification stage corresponding to the CBCT sample of the permanent anterior teeth. The weights and biases of the trained convolutional neural network are adjusted using the validation set to obtain the adjusted convolutional neural network. The adjusted convolutional neural network was used as the prediction model for the permanent anterior teeth.

2. The method for predicting permanent tooth development according to claim 1, characterized in that, The oral CBCT scans to be staged are segmented to obtain oral CBCT scans of different teeth, specifically including: The open-source annotation tool 3D Slicer was used to segment the oral CBCT of the teeth to be staged, resulting in oral CBCT of different teeth.

3. The method for predicting permanent tooth development according to claim 1, characterized in that, The tooth position representation methods include: FDI tooth position representation, location recording method, and general recording method.

4. A system for predicting permanent tooth development, characterized in that, The system for predicting permanent tooth development is applied to the method for predicting permanent tooth development according to any one of claims 1-3, and the system for predicting permanent tooth development comprises: The acquisition module is used to acquire the oral CBCT scan to be staged. The segmentation module is used to segment the oral CBCT to be staged, resulting in oral CBCTs of different teeth. The encoding module is used to encode the oral CBCT of different teeth using tooth position representation, so as to obtain the oral CBCT after encoding different teeth; The classification module is used to determine the permanent anterior teeth, permanent premolars, and permanent molars to be staged based on the oral CBCT scans of different teeth and the root morphology of different teeth. The permanent anterior tooth prediction module is used to input the permanent anterior teeth to be staged into the permanent anterior tooth prediction model to obtain the calcification stage of the permanent anterior teeth. The permanent premolar prediction module is used to input the permanent premolar to be staged into the permanent premolar prediction model to obtain the calcification stage of the permanent premolar. The permanent molar prediction module is used to input the permanent molar to be staged into the permanent molar prediction model to obtain the calcification stage of the permanent molar; The training module specifically includes: Building units are used to construct convolutional neural networks; The acquisition unit is used to acquire CBCT training data of the permanent anterior teeth oral cavity; the CBCT training data of the permanent anterior teeth oral cavity includes CBCT samples of the permanent anterior teeth oral cavity and the calcification stage corresponding to the CBCT samples of the permanent anterior teeth oral cavity. A classification unit is used to divide the permanent anterior dentist oral CBCT training data into a training set and a validation set; The training unit is used to train the weights and biases of the convolutional neural network using the training set with the aim of minimizing the loss function, so as to obtain the trained convolutional neural network; the loss function is determined based on the error between the output of the convolutional neural network and the calcification stage corresponding to the CBCT sample of the permanent anterior teeth. The verification unit is used to adjust the weights and biases of the trained convolutional neural network using the verification set to obtain the adjusted convolutional neural network. The output unit is used to use the adjusted convolutional neural network as the prediction model for the permanent anterior teeth.

5. An electronic device, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the method as described in any one of claims 1 to 3.

6. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, which, when executed, implements the method as described in any one of claims 1 to 3.

Images