Local alveolar bone defect automatic identification method, device, equipment, medium and product

Abstract

The invention discloses a local alveolar bone defect automatic identification method and device, equipment, a medium and a product, and relates to the field of oral cavity, and the method comprises the steps: obtaining a picture corresponding to a target tooth position; the pictures corresponding to the target tooth position are N continuous target sagittal plane slices on the target CBCT image; the target CBCT image is a CBCT scanning result containing oral and maxillofacial region image information of the patient; the N continuous target sagittal plane slices are all slices comprising teeth and alveolar bones corresponding to the target tooth positions. Inputting a picture corresponding to the target tooth position into the local alveolar bone defect recognition model to obtain a recognition result; the local alveolar bone defect recognition model is obtained by training a deep learning model. The method can reduce the workload of doctors, improve the diagnosis efficiency, objectively provide reference for the doctors, provide auxiliary diagnosis for clinicians, reduce diagnosis deviation possibly caused by manual naked eye judgment, and improve the diagnosis accuracy.

Description

Technical Field

[0001] The present application relates to the oral field, and in particular to a method, device, equipment, medium and product for automatically identifying local alveolar bone defects. Background Art

[0002] During orthodontic treatment, if the tooth root exceeds the limit of the periodontal supporting tissue, alveolar bone defects (ABD) may occur. Common alveolar bone defects include bone fenestration and bone cracking. Alveolar bone defects not only affect the patient's oral health, but also increase the difficulty of orthodontic treatment and even affect the effect of orthodontic treatment. Therefore, in order to avoid the formation of alveolar bone defects during orthodontic treatment, the doctor must determine the morphology of the patient's alveolar bone before orthodontic treatment, so that a suitable orthodontic plan can be formulated while ensuring the patient's periodontal health; during orthodontic treatment, if a large degree of movement is required, corresponding measures must be taken to observe the condition of the alveolar bone.

[0003] Typically, there are two methods for examining alveolar bone defects such as bone fenestrations and bone dehiscence. One is direct observation through periodontal flap surgery, considered the gold standard. However, this invasive approach is difficult for patients to accept. Therefore, another method is often chosen clinically: cone beam computed tomography (CBCT) imaging to detect alveolar bone defects. Compared with periodontal flap surgery, CBCT offers the advantages of low risk, convenience, and efficiency. CBCT has high accuracy and reliability in detecting these defects.

[0004] However, this method also requires doctors to rely on professional knowledge and clinical experience to analyze CBCT, which is highly subjective and involves a huge workload. Summary of the Invention

[0005] The purpose of this application is to provide a method, device, equipment, medium and product for automatic identification of local alveolar bone defects, which can reduce the workload of doctors, improve diagnostic efficiency, and objectively provide reference for doctors, provide auxiliary diagnosis for clinical doctors, reduce diagnostic deviations that may be caused by manual visual judgment, and improve diagnostic accuracy.

[0006] To achieve the above objectives, this application provides the following solutions:

[0007] In a first aspect, the present application provides a method for automatically identifying a local alveolar bone defect, comprising:

[0008] Obtain an image corresponding to the target tooth position; the image corresponding to the target tooth position is N consecutive target sagittal slices on the target CBCT image, where N is an integer greater than or equal to 3; the target CBCT image is a CBCT scan result containing oral and maxillofacial imaging information of the patient; and the N consecutive target sagittal slices all include slices of the tooth and alveolar bone corresponding to the target tooth position;

[0009] The image corresponding to the target tooth position is input into the local alveolar bone defect recognition model to obtain a recognition result; the recognition result is whether the alveolar bone corresponding to the target tooth position has a defect; the local alveolar bone defect recognition model is obtained by training a deep learning model.

[0010] In one embodiment, the deep learning model includes: a feature extraction module, a feature splicing module and a fully connected layer connected in sequence; the feature extraction module includes N parallel convolutional neural networks.

[0011] In one embodiment, the deep learning model further includes: an adaptive average pooling layer; the feature splicing module and the fully connected layer are connected through the adaptive average pooling layer.

[0012] In one embodiment, the process of training the deep learning model to obtain the local alveolar bone defect recognition model specifically includes:

[0013] Obtain a sample set; the sample set includes an image and a label corresponding to each sample tooth position; the label is whether the alveolar bone is defective or not defective; the image corresponding to the sample tooth position is N consecutive sample sagittal slices on the sample CBCT image; the sample CBCT image is a CBCT scan result containing oral and maxillofacial imaging information of the sample patient; the N consecutive sample sagittal slices all include slices of the tooth and alveolar bone corresponding to the sample tooth position;

[0014] Dividing the sample set into a training set and a test set;

[0015] Taking the image corresponding to the sample tooth position as input and the recognition result of the sample tooth position as output, a five-fold cross-validation method is adopted to train the deep learning model according to the training set to obtain a trained model;

[0016] The trained model is tested using the test set to obtain a local alveolar bone defect recognition model.

[0017] In one embodiment, the convolutional neural network is ResNet50.

[0018] In one embodiment, the process of determining the sample sagittal slice on the sample CBCT image is as follows:

[0019] The original sagittal slices on the sample CBCT image are sequentially subjected to region of interest cropping and image enhancement processing to obtain sample sagittal slices on the sample CBCT image.

[0020] In a second aspect, the present application provides an automatic identification device for partial alveolar bone defects, comprising:

[0021] an acquisition module, configured to acquire an image corresponding to a target tooth position; the image corresponding to the target tooth position being N consecutive target sagittal slices on a target CBCT image, where N is an integer greater than or equal to 3; the target CBCT image being a CBCT scan result containing oral and maxillofacial imaging information of the patient; and the N consecutive target sagittal slices all being slices including the tooth and alveolar bone corresponding to the target tooth position;

[0022] The recognition module is used to input the image corresponding to the target tooth position into the local alveolar bone defect recognition model to obtain a recognition result; the recognition result is whether the alveolar bone corresponding to the target tooth position has a defect; the local alveolar bone defect recognition model is obtained by training a deep learning model.

[0023] In a third aspect, the present application provides a computer device comprising: 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 any of the above-described methods for automatically identifying local alveolar bone defects.

[0024] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the above-mentioned methods for automatically identifying local alveolar bone defects.

[0025] In a fifth aspect, the present application provides a computer program product, comprising a computer program, which, when executed by a processor, implements any of the above-mentioned methods for automatically identifying local alveolar bone defects.

[0026] According to the specific embodiments provided in this application, this application has the following technical effects:

[0027] The present application provides a method, device, equipment, medium, and product for automatically identifying local alveolar bone defects. Currently, most studies have shown that CBCT examinations for alveolar bone defects are highly accurate and reliable. However, in clinical practice, doctors rely on their professional knowledge and clinical experience to inevitably be subjective when analyzing whether an alveolar bone defect has occurred. Furthermore, dentists need to analyze each patient's condition when performing orthodontic work, which is a huge workload, and the various measurements and analyses are time-consuming for both doctors and patients. The present application automatically obtains recognition results by inputting images corresponding to the target tooth position into a local alveolar bone defect recognition model, replacing manual labor to reduce the doctor's workload and improve diagnostic efficiency. Furthermore, the model-based recognition method can produce objective recognition results, providing doctors with objective reference and assisting in diagnosis. This reduces diagnostic bias that may result from manual visual judgment, improves diagnostic accuracy, and enables doctors to better formulate or modify orthodontic plans to ensure patient treatment outcomes. It also saves patients time during their visits. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0029] Figure 1 This is a diagram of an application environment of a method for automatically identifying a local alveolar bone defect in an embodiment of the present application;

[0030] Figure 2 Schematic diagram of the alveolar bone condition;

[0031] Figure 3 Schematic diagram of the deep learning model training process;

[0032] Figure 4 A schematic diagram of the functional modules of an automatic identification device for partial alveolar bone defects provided in one embodiment of the present application;

[0033] Figure 5 A schematic diagram of a specific process flow of a method for automatically identifying a local alveolar bone defect provided in one embodiment of the present application;

[0034] Figure 6 This is a diagram of experimental results provided in one embodiment of the present application;

[0035] Figure 7 A schematic diagram of a confusion matrix provided in one embodiment of the present application;

[0036] Figure 8A schematic diagram of the general flow of a method for automatically identifying a local alveolar bone defect provided in one embodiment of the present application;

[0037] Figure 9 A schematic diagram of the structure of a computer device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0038] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0039] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0040] The automatic identification method of local alveolar bone defect provided in the embodiment of the present application can be applied to Figure 1 In the application environment shown. Among them, the terminal 102 communicates with the server 104 through the network. The data storage system can store the data that the server 104 needs to process. The data storage system can be set up separately, integrated on the server 104, or placed on the cloud or other servers. The terminal 102 can send the picture corresponding to the target tooth position to the server 104. After the server 104 receives the picture corresponding to the target tooth position, the server 104 inputs the picture corresponding to the target tooth position into the local alveolar bone defect recognition model to obtain a recognition result; the recognition result is whether the alveolar bone of the target tooth position is defective; the local alveolar bone defect recognition model is obtained by training the deep learning model. The server 104 can feedback the obtained defect result to the terminal 102. In addition, in some embodiments, the method for automatic identification of local alveolar bone defects can also be implemented independently by the server 104 or the terminal 102. For example, the terminal 102 can directly perform automatic identification of local alveolar bone defects on the picture corresponding to the target tooth position, or the server 104 can obtain the picture corresponding to the target tooth position from the data storage system and perform automatic identification of local alveolar bone defects on the picture corresponding to the target tooth position.

[0041] In an exemplary embodiment, Figure 5 and Figure 8 As shown, a method for automatically identifying a local alveolar bone defect is provided. The method is executed by a computer device, specifically a computer device such as a terminal or a server, or a terminal and a server. In the embodiment of the present application, the method is applied to Figure 1The server 104 in FIG. 1 is used as an example to illustrate the method, which includes the following steps, wherein:

[0042] Step 201: Obtain an image corresponding to the target tooth position; the image corresponding to the target tooth position is N consecutive target sagittal slices on the target CBCT image, where N is an integer greater than or equal to 3; the target CBCT image is a CBCT scan result containing the patient's oral and maxillofacial imaging information; the N consecutive target sagittal slices all include slices of the tooth and alveolar bone corresponding to the target tooth position.

[0043] Step 202: Input the image corresponding to the target tooth position into the local alveolar bone defect recognition model to obtain a recognition result; the recognition result is whether the alveolar bone corresponding to the target tooth position has a defect; the local alveolar bone defect recognition model is obtained by training a deep learning model.

[0044] In another exemplary embodiment of the present application, the process of training the deep learning model to obtain a local alveolar bone defect recognition model specifically includes:

[0045] A sample set is obtained; the sample set includes images and labels corresponding to each sample tooth position; the label is whether the alveolar bone is defective or not defective; the images corresponding to the sample tooth position are N consecutive sample sagittal slices on the sample CBCT image; the sample CBCT image is a CBCT scan result containing oral and maxillofacial imaging information of the sample patient; the N consecutive sample sagittal slices all include slices of the teeth and alveolar bone corresponding to the sample tooth position.

[0046] The sample set is divided into a training set and a test set. Specifically, the sample set can be divided into a training set and a test set in a ratio of 8:2.

[0047] The image corresponding to the sample tooth position is used as input and the recognition result of the sample tooth position is used as output. The five-fold cross-validation method is adopted to train the deep learning model according to the training set to obtain a trained model. During training, the epoch is 90, the batch size is 16, the initial learning rate is 1e-5, the weight decay ratio is 5e-2, and the adaptive moment estimation (ADAM) optimization algorithm is used to reduce the loss function value and update the network weight parameters.

[0048] The trained model is tested using the test set to obtain a local alveolar bone defect recognition model.

[0049] In another exemplary embodiment of the present application, the process for determining sagittal slices on a sample CBCT image is as follows: The image reading software is launched, and the sample CBCT images are sequentially opened in a panoramic view. The sagittal display interface is adjusted to a 1×3 format, and the slice thickness is adjusted to 0.4 mm. The three displayed original sagittal slices are saved in sequence. Subsequently, the original sagittal slices are sequentially cropped to a region of interest and image enhanced to obtain sagittal slices.

[0050] In another exemplary embodiment of the present application, obtaining a sample set specifically includes:

[0051] Step 1: Screen eligible cases (alveolar bone defect) from the case database and obtain their CBCT images.

[0052] All the teeth of patients with alveolar bone defects (ABD) were identified by reviewing the electronic medical records. At the same time, measurements were performed to determine if the defects had the following conditions on the CBCT images: (Specifically, if the distance between the enamel-cementum junction and the deepest point of the bone defect was greater than 2 mm, it was considered a bone crack. If the bone defect did not include the alveolar ridge top and its vertical measurement was greater than 0 mm, it was classified as a bone fenestration) and the original sagittal slices of the alveolar bone defect teeth were captured on the panoramic view of the CBCT image, such as Figure 2 As shown, Figure 2 The three pictures in part (a) are examples when the alveolar bone is not missing; Figure 2 Parts (b) and (c) are examples of alveolar bone defects. Figure 2 The three pictures in the middle (b) are all bone fractures. Figure 2 The middle (c) part is a bone window. Since alveolar bone defect is defined as any absence of cortical bone around the vestibular surface of the root in at least three consecutive sagittal views, at least three consecutive original sagittal slices of each tooth position need to be collected. Figure 2 Parts (a), (b), and (c) all contain three consecutive sagittal slices of the target tooth position. Place at least three consecutive slices of each tooth position in the same subfolder.

[0053] Step 2: Obtain the original sagittal slices of the tooth with no alveolar bone defect (Non-ABD).

[0054] For patients with intact alveolar bone, all teeth from the central incisor to the first premolar were first included, and then measurements were performed on the CBCT images (the distance between the cementoenamel junction and the alveolar ridge top was measured and recorded using a linear measurement tool on the three sagittal slices displayed). Based on the measurement results and the range described in step 1, the teeth that did not meet the requirements for intact alveolar bone were excluded to determine the teeth with intact alveolar bone. At least three consecutive original sagittal slices of each tooth with intact alveolar bone were captured, as shown in Figure 2. Figure 2 As shown in part (a), place at least three consecutive slices of each tooth position in the same subfolder.

[0055] Step 3: Classify the above subfolders according to tags. Each tag corresponds to a parent folder. All the above subfolders are placed in the corresponding parent folder according to the tags.

[0056] Step 4: Crop the region of interest for all slices to reduce the interference of irrelevant parts of the image on the training; then use Gaussian sharpening to enhance the original image to improve image blur.

[0057] This application selects at least three slices to be input into the network for training, and finally obtains the corresponding judgment. For the multi-image input form of this application, a custom data set class MyDatasets is designed for loading and preprocessing image data. This class inherits from the Dataset class of PyTorch and rewrites the _getitem_ method to realize the loading, preprocessing and indexing functions of data. First, two categories (ABD, Non-ABD) are determined by the parent folder name of the subfolder to generate labels. Secondly, data loading is performed, the three images in the subfolder are read, and finally data enhancement is performed.

[0058] The deep learning model selects ResNet50 in the convolutional neural network. Since the original ResNet50 only accepts a single input image, in order to allow the subfolder containing at least three slices to be input into the network, the selected network is modified to a certain extent. Based on ResNet50, it is expanded to a multi-input model for fusing multiple image features, and the ability to support multiple image inputs is expanded. The features of each image are extracted by calling the ResNet50 backbone network three times, and then the feature vectors of the three images are spliced ​​along the channel dimension. An adaptive average pooling module is added to the spliced ​​features in order to compress the high-dimensional features to a fixed size, so as to adapt to the subsequent full connection operation of the multi-input spliced ​​features, and finally the spliced ​​features are mapped to the classification space through a fully connected layer. For the fully connected layer, after the three image features are spliced, the number of input channels becomes 2048×3, and the final fully connected layer size also changes accordingly. According to the binary classification task of this application, the output layer is changed from the original 1000 to 2. Therefore, in another exemplary embodiment of the present application, such as Figure 3 As shown, the deep learning model includes: a feature extraction module, a feature splicing module and a fully connected layer connected in sequence; the feature extraction module includes N parallel convolutional neural networks.

[0059] In another exemplary embodiment of the present application, the deep learning model further includes an adaptive average pooling layer, wherein the feature concatenation module and the fully connected layer are connected via the adaptive average pooling layer. The adaptive average pooling layer is used to compress high-dimensional features to a fixed size to accommodate subsequent fully connected operations on multiple input concatenated features.

[0060] In another exemplary embodiment of the present application, the convolutional neural network is ResNet50.

[0061] In another exemplary embodiment of the present application, the trained model is tested using the test set to obtain a localized alveolar bone defect recognition model. Specifically, the model is tested on the test set, and the diagnostic performance of the model is evaluated by calculating relevant indicators such as accuracy, precision, specificity, recall rate, F1 score, and AUC.

[0062] This application also provides an embodiment to verify the effectiveness of the automatic identification method for local alveolar bone defects provided in this application. The specific steps are as follows:

[0063] Dataset Preparation: We collected matching CBCT images based on the inclusion criteria and determined the specific tooth positions. Three consecutive sagittal slices were selected for each tooth position and placed in the same subfolder. This meant that each category contained several subfolders containing three slices each for training and validation. Finally, the binary classification dataset was randomly partitioned into a training set:test set ratio of 8:2.

[0064] Model Training: Following the aforementioned model design and data structure, the ResNet50 network was modified to allow for smooth input of multiple slices for training. To better utilize the dataset, training was performed using a five-fold cross-validation method. Five iterations were performed, with one subset selected as the validation set in each iteration and the remaining subsets combined as the training set. Finally, training accuracy curves for each fold were generated to visualize the effectiveness of model training.

[0065] Model evaluation: Test the model on the test set and first get the confusion matrix to visualize and compare the classification results. Figure 7 The results are shown in Table 1 to evaluate the diagnostic performance of the model. Figure 6 As shown, Figure 6 Parts (a), (b), (c), (d), and (e) are the accuracy curves of the training set and validation set for the first, second, third, fourth, and fifth folds in the five-fold cross-validation, respectively.

[0066] Table 1 Model test results

[0067] Model Accuracy Precision Specificity Recall F1Score AUC ResNet50 0.912 0.989 0.994 0.792 0.878 0.984

[0068] This application not only explores the effect of ResNet50 as the base model in the automatic identification of local alveolar bone defects, but also compares it with the ResNet34 and ResNet152 models of the ResNet series. The ResNet50 in the deep learning model provided in the embodiment of this application is replaced with the ResNet34 and ResNet152 models, and then trained using the same method as the above embodiment. The diagnostic performance of the model is evaluated on the test set. The test results are shown in Table 2. ResNet50 shows the best performance under the 5-fold cross-validation method.

[0069] Table 2 Test results of each model

[0070] Accuracy Precision Specificity Recall F1-score AUC ResNet34 0.870 0.965 0.983 0.705 0.813 0.979 ResNet50 0.912 0.989 0.994 0.792 0.878 0.984 ResNet152 0.874 0.958 0.977 0.722 0.820 0.962

[0071] This application proposes for the first time the automatic identification of local alveolar bone defects on CBCT images through convolutional neural networks. This method can reduce the workload of doctors in clinical practice and effectively improve the diagnosis speed of orthodontists. At the same time, it can objectively provide clinicians with diagnostic results, and provide diagnostic references for inexperienced clinicians, so that clinicians can better formulate or promptly revise orthodontic plans based on the alveolar bone conditions, ensure the orthodontic treatment effect of patients, and solve the problems of time-consuming, highly subjective and decision-making uncertainty in manual reading of films, and prevent diagnostic deviations that may be caused by manual visual judgment from affecting the orthodontic effect of patients.

[0072] Based on the same inventive concept, the embodiments of the present application also provide a device for automatically identifying a partial alveolar bone defect for implementing the aforementioned method for automatically identifying a partial alveolar bone defect. The solution provided by the device is similar to the solution described in the aforementioned method. Therefore, the specific limitations of one or more embodiments of the device for automatically identifying a partial alveolar bone defect provided below can be found in the above-mentioned limitations of the method for automatically identifying a partial alveolar bone defect, and will not be further elaborated here.

[0073] In an exemplary embodiment, Figure 4 As shown, a device for automatically identifying local alveolar bone defects is provided, comprising:

[0074] Acquisition module A1 is used to acquire an image corresponding to the target tooth position; the image corresponding to the target tooth position is N consecutive target sagittal slices on the target CBCT image, where N is an integer greater than or equal to 3; the target CBCT image is a CBCT scan result containing the patient's oral and maxillofacial imaging information; the N consecutive target sagittal slices all include slices of the tooth and alveolar bone corresponding to the target tooth position.

[0075] Identification module A2 is used to input the picture corresponding to the target tooth position into the local alveolar bone defect recognition model to obtain a recognition result; the recognition result is whether the alveolar bone corresponding to the target tooth position has a defect; the local alveolar bone defect recognition model is obtained by training a deep learning model.

[0076] In an exemplary embodiment, a computer device is provided. The computer device may be a server or a terminal. The internal structure diagram thereof may be as follows: Figure 9As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, abbreviated as I / O) and a communication interface. The processor, memory and input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The database of the computer device is used to store local alveolar bone defect automatic identification data. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a method for automatic identification of local alveolar bone defects is implemented.

[0077] Those skilled in the art will understand that Figure 9 The structure shown in the figure is merely a block diagram of a portion of the structure related to the solution of the present application and does not constitute a limitation on the computer device to which the solution of the present application is applied. A specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement. In an exemplary embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and the processor implements the above-mentioned method embodiments when executing the computer program.

[0078] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, which implements the above-mentioned method embodiments when executed by a processor.

[0079] In an exemplary embodiment, a computer program product is provided, including a computer program. When the computer program is executed by a processor, the above method embodiments are implemented.

[0080] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0081] Those skilled in the art will understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, database or other media used in the embodiments provided in this application may include at least one of non-volatile and volatile memory. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory may include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0082] The databases involved in the various embodiments provided herein may include at least one of a relational database and a non-relational database. Non-relational databases may include, but are not limited to, distributed databases based on blockchains. The processors involved in the various embodiments provided herein may include, but are not limited to, general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic units, data processing logic units based on quantum computing, and the like.

[0083] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0084] This document uses specific examples to illustrate the principles and implementation methods of this application. The description of the above examples is only intended to help understand the method and core concept of this application. At the same time, for those skilled in the art, based on the concept of this application, there may be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A method for automatically identifying local alveolar bone defects, characterized in that: The method for automatically identifying local alveolar bone defects comprises: Obtain an image corresponding to the target tooth position; the image corresponding to the target tooth position is N consecutive target sagittal slices on the target CBCT image, where N is an integer greater than or equal to 3; the target CBCT image is a CBCT scan result containing oral and maxillofacial imaging information of the patient; and the N consecutive target sagittal slices all include slices of the tooth and alveolar bone corresponding to the target tooth position; The image corresponding to the target tooth position is input into the local alveolar bone defect recognition model to obtain a recognition result; the recognition result is whether the alveolar bone corresponding to the target tooth position has a defect; the local alveolar bone defect recognition model is obtained by training a deep learning model.

2. The automatic identification method for partial alveolar bone defects according to claim 1, characterized in that: The deep learning model includes: a feature extraction module, a feature splicing module and a fully connected layer connected in sequence; the feature extraction module includes N parallel convolutional neural networks.

3. The automatic identification method for partial alveolar bone defect according to claim 2, characterized in that: The deep learning model also includes: Adaptive average pooling layer; the feature splicing module and the fully connected layer are connected through the adaptive average pooling layer.

4. The automatic identification method for partial alveolar bone defects according to claim 2, characterized in that: The process of training the deep learning model to obtain a local alveolar bone defect recognition model specifically includes: Obtain a sample set; the sample set includes an image and a label corresponding to each sample tooth position; the label is whether the alveolar bone is defective or not defective; the image corresponding to the sample tooth position is N consecutive sample sagittal slices on the sample CBCT image; the sample CBCT image is a CBCT scan result containing oral and maxillofacial imaging information of the sample patient; the N consecutive sample sagittal slices all include slices of the tooth and alveolar bone corresponding to the sample tooth position; Dividing the sample set into a training set and a test set; Taking the image corresponding to the sample tooth position as input and the recognition result of the sample tooth position as output, a five-fold cross-validation method is adopted to train the deep learning model according to the training set to obtain a trained model; The trained model is tested using the test set to obtain a local alveolar bone defect recognition model.

5. The automatic identification method for partial alveolar bone defect according to claim 2, characterized in that: The convolutional neural network is ResNet50.

6. The automatic identification method for partial alveolar bone defects according to claim 4, characterized in that: The process of determining the sample sagittal slice on the sample CBCT image is as follows: The original sagittal slices on the sample CBCT image are sequentially subjected to region of interest cropping and image enhancement processing to obtain sample sagittal slices on the sample CBCT image.

7. An automatic identification device for partial alveolar bone defects, characterized in that: The automatic identification device for local alveolar bone defects comprises: an acquisition module, configured to acquire an image corresponding to a target tooth position; the image corresponding to the target tooth position being N consecutive target sagittal slices on a target CBCT image, where N is an integer greater than or equal to 3; the target CBCT image being a CBCT scan result containing oral and maxillofacial imaging information of the patient; and the N consecutive target sagittal slices all being slices including the tooth and alveolar bone corresponding to the target tooth position; The recognition module is used to input the image corresponding to the target tooth position into the local alveolar bone defect recognition model to obtain a recognition result; the recognition result is whether the alveolar bone corresponding to the target tooth position has a defect; the local alveolar bone defect recognition model is obtained by training a deep learning model.

8. A computer device comprising: 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 automatically identifying a local alveolar bone defect according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for automatically identifying a partial alveolar bone defect according to any one of claims 1 to 6 is implemented.

10. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the method for automatically identifying a partial alveolar bone defect according to any one of claims 1 to 6 is implemented.

Images