Dental image-based decayed tooth identification analysis early warning method and system

By performing feature extraction and deep learning analysis on dental images, combined with recurrent neural networks, the problem of difficult to identify early caries and assess caries risks in the existing technology is solved, personalized caries risk assessment and early warning, and improved dental health management accuracy and preventive effect.

CN118983086AInactive Publication Date: 2024-11-19ZHEJIANG CHINESE MEDICAL UNIVERSITY
View PDF 0 Cites 4 Cited by

Patent Information

Application Number
CN202410977391.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-19
Publication Date
2024-11-19
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The existing caries identification methods are difficult to detect early caries in a timely manner, and the caries risk assessment is not accurate enough to effectively prevent and deal with caries, and there is a lack of predictions on future caries risks.

Method used

By acquiring dental images, extracting dental data, generating dental feature vectors, using deep learning methods and caries feature threshold analysis, marking caries risk, and using recurrent neural networks for timing analysis, predicting future caries risk and generating caries risk detection reports.

Benefits of technology

It has achieved sensitive identification of early caries, provided personalized caries risk assessment and real-time warning, reduced the risk of caries progress, provided detailed caries risk detection reports and prevention suggestions, and improved the accuracy and effectiveness of dental health management.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118983086A_ABST
    Figure CN118983086A_ABST
Patent Text Reader

Abstract

The invention discloses a dental caries identification analysis early warning method and system based on a tooth image, and relates to the technical field of dental caries identification analysis early warning. User tooth data is extracted through a user tooth image to generate a user tooth feature vector value, and the user tooth feature vector value and a dental caries feature vector threshold value are used for analysis; marking and evaluating whether the user has a dental caries risk; further identifying a decayed tooth risk grade according to the marked condition of the decayed tooth risk of the user, and giving an alarm according to the decayed tooth risk grade; and for the condition that the user does not have the dental caries risk, obtaining historical tooth data of the user, performing time sequence analysis on the historical tooth data of the user by using a recurrent neural network, predicting the probability of the dental caries risk of the user, and finally generating a dental caries risk detection report of the user. Personalized dental caries risk assessment and real-time early warning are realized, and comprehensive dental caries management suggestions are provided through historical data analysis and prediction.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of caries identification, analysis and early warning, and in particular to a caries identification, analysis and early warning method and system based on tooth images. Background Art

[0002] Oral health has an important impact on overall health. Dental caries is one of the most common oral diseases. Early detection and effective management of dental caries are important for preventing tooth damage and reducing the need for dental treatment.

[0003] Traditional methods for identifying caries often rely on the clinical experience of dentists for caries diagnosis. However, since the symptoms of early caries are not obvious and it is difficult for non-professionals to accurately judge, early caries (especially shallow caries) are difficult to be discovered in time. In addition, the traditional caries detection process may require users to go to hospitals or clinics frequently, which is not only time-consuming and labor-intensive, but may also increase the financial burden on users. With the rapid development of computer vision and image processing technology, especially the rise of deep learning methods, researchers have begun to explore how to use these technologies to solve problems in the medical field. However, existing caries risk assessment methods are not accurate enough to accurately assess the user's caries risk level, and thus cannot provide effective prevention and response measures. In addition, existing methods can often only assess the current dental condition, but cannot predict the caries risk that may occur in the future. At the same time, some existing caries detection methods still require a lot of manual intervention and judgment, and the degree of automation and intelligence is not high.

[0004] Therefore, in response to the above problems, there is an urgent need for a caries identification, analysis and early warning method and system based on tooth images. Summary of the invention

[0005] In view of the shortcomings of the existing technology, the present invention provides a caries identification, analysis and early warning method and system based on tooth images, which solves the problem that early caries are difficult to detect in time, and the existing caries risk assessment methods are not precise enough to accurately assess the user's caries risk level, and thus cannot provide effective prevention and response measures, and lack the prediction of caries risks that may occur in the future.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: a dental caries identification, analysis and early warning method based on dental images, comprising the following steps: obtaining a user's dental image, and then extracting the user's dental data through the user's dental image; generating a user's dental feature vector value based on the user's dental data, and using the user's dental feature vector value and the dental caries feature vector threshold for analysis to mark and evaluate whether the user has a dental caries risk; further identifying the dental caries risk level for the marked user who has a dental caries risk, and issuing an alarm based on the dental caries risk level; for the user who does not have a dental caries risk, obtaining the user's historical dental data, and using a recurrent neural network to perform time series analysis on the user's historical dental data to predict the probability of the user having a dental caries risk; and generating a dental caries risk detection report for the user.

[0007] Furthermore, the specific analysis of extracting user tooth data through user tooth image is as follows: preprocessing the user tooth image to obtain a preprocessed user tooth image; the preprocessing includes adjusting image brightness and contrast, image denoising, and image size normalization operations; locating and segmenting the tooth area in the preprocessed user tooth image based on the image segmentation algorithm, and determining the position and boundary of each tooth to obtain a segmented user tooth image; using a feature extraction method to extract user tooth data from the segmented user tooth image, the user tooth data specifically includes morphological data of each tooth, color data of each tooth, and texture data of each tooth; the morphological data of each tooth is specifically the tooth surface convexity, the color data of each tooth is specifically the tooth surface grayscale value, and the texture data of each tooth is specifically the plaque extension range.

[0008] Furthermore, the specific analysis of generating the user's tooth feature vector value based on the user's tooth data is: numbering each tooth of the user, combining the tooth surface convexity, tooth surface grayscale value and dental plaque extension range of each tooth into the feature vector of each tooth; for the feature vector of each tooth, training is performed using a support vector machine to output the feature vector value of each tooth.

[0009] Furthermore, the specific analysis of using the user's tooth feature vector value and the caries feature vector threshold to analyze and mark and evaluate whether the user has a caries risk is as follows: a computer vision method based on deep learning is used to classify the user's teeth into incisors, canines, premolars and molars, and the caries feature vector thresholds of the incisors, the caries feature vector thresholds of the canines, the caries feature vector thresholds of the premolars and the caries feature vector thresholds of the molars are obtained respectively; the feature vector values ​​of each tooth of the user are compared with the caries feature vector thresholds of different classifications according to the user's tooth classification, and teeth with feature vector values ​​greater than or equal to the caries feature vector threshold are marked as teeth with caries risk, and when a marking situation exists, the user is assessed to have a caries risk.

[0010] Furthermore, the specific analysis for further identifying the caries risk level of the marked user with caries risk is as follows: the total number of marks is obtained for the marked user with caries risk, and then the teeth at each marked position are renumbered; the Euclidean distance calculation is performed on the eigenvector values ​​of the re-numbered teeth and the caries eigenvector thresholds corresponding to different classifications, and then the Euclidean distance calculation results of all the re-numbered teeth are averaged to obtain the user's caries risk value, and the user's caries risk level is identified based on the user's caries risk value.

[0011] Furthermore, in view of the situation that the user does not have the risk of dental caries, the user's historical dental data is obtained, and the user's historical dental data is analyzed in time series using a recurrent neural network. The specific analysis for predicting the probability of the user having the risk of dental caries is as follows: the user's historical dental data includes the tooth surface convexity and concavity, tooth surface grayscale value and dental plaque extension range extracted from the user's historically uploaded tooth images; the tooth surface convexity and concavity, tooth surface grayscale value and dental plaque extension range are extracted from the user's historically uploaded tooth images for training to obtain a trained recurrent neural network; the user's latest dental data is obtained, the user's latest dental data is used as the input of the trained recurrent neural network, the user's latest dental data is predicted, and the prediction result, i.e., the probability of the risk of dental caries at the next time point, is output.

[0012] Furthermore, the specific training method of the recurrent neural network is as follows: the tooth surface convexity and concavity, tooth surface grayscale value and dental plaque extension range extracted from the tooth images uploaded by the user in the past are combined into data sets at each time point in chronological order to form a time series, and then the data sets are divided into training set, validation set and test set; the hierarchical structure of the recurrent neural network specifically includes: the input layer is used to receive historical tooth data, the hidden layer is used to capture the dependency relationship in the time series, and the output layer is used to predict the probability of caries risk at the next time point; the loss function and optimizer are selected, and then the training set is used for model training, the model parameters are adjusted through the validation set, and the model performance is evaluated using the test set. The model performance includes accuracy, precision, recall rate and F1 score.

[0013] Furthermore, the caries risk detection report specifically includes whether the user is at risk of caries, the mark location where the user is at risk of caries, the user's caries risk level, the probability of the user having caries risk, and caries coping suggestions; caries coping suggestions specifically include coping suggestions when the user is at risk of caries and coping suggestions when the user is not at risk of caries; coping suggestions when the user is at risk of caries specifically include medical treatment suggestions, dietary suggestions, and brushing method suggestions; coping suggestions when the user is not at risk of caries specifically include prevention suggestions, dietary suggestions, and brushing method suggestions.

[0014] A dental caries recognition, analysis and early warning system based on dental images, which applies the above-mentioned dental caries recognition, analysis and early warning method based on dental images, includes: a data acquisition module, which is used to acquire a user's dental image, and then extract the user's dental data through the user's dental image; a dental caries risk identification module, which is used to generate a user's dental feature vector value based on the user's dental data, and use the user's dental feature vector value and the dental caries feature vector threshold for analysis to mark and evaluate whether the user has a dental caries risk; a risk level identification module, which is used to further identify the dental caries risk level for the marked user with a dental caries risk, and issue an alarm based on the dental caries risk level; a risk prediction module, which is used to acquire the user's historical dental data for the situation where the user does not have a dental caries risk, and use a recurrent neural network to perform time series analysis on the user's historical dental data to predict the probability of the user having a dental caries risk; and a report generation module, which is used to generate a dental caries risk detection report for the user.

[0015] The present invention has the following beneficial effects:

[0016] The dental caries identification, analysis and early warning method and system based on dental images can effectively identify early signs of dental caries by acquiring the user's dental images and extracting dental data, combined with the analysis of dental caries feature vectors. Early dental caries may not be easily detected during naked eye inspection, but through image processing and feature extraction technology, tiny dental caries lesions can be captured, so that preventive measures can be taken in time to prevent the disease from worsening. By generating the user's dental feature vector values ​​and comparing them with thresholds, a personalized assessment of the dental caries risk of each user can be performed. The personalized assessment takes into account the user's specific dental conditions and characteristics, and is more accurate and reliable than traditional universal assessment methods. For users marked as having dental caries risk, the level of dental caries risk can be further identified. , and issue an alarm in time. The real-time early warning mechanism helps dentists or users themselves take timely measures, such as regular check-ups, changing eating habits or enhancing oral hygiene measures, thereby reducing the risk of dental caries progression; for users who are not at risk of dental caries, time series analysis using recurrent neural networks can predict the probability of dental caries risk in the future, eliminate potential risks during the initial assessment, and monitor the dynamic changes in dental caries risk, providing more comprehensive and long-term prevention recommendations; the dental caries risk detection report not only informs users whether there is a dental caries risk, but also lists in detail the risk location, dental caries risk level, risk probability and response recommendations, which helps users understand their own oral health status and take appropriate preventive measures to maintain dental health. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 This is a flow chart of the caries identification, analysis and early warning method based on tooth images of the present invention.

[0018] Figure 2This is a structural diagram of the caries identification, analysis and early warning system based on tooth images of the present invention. DETAILED DESCRIPTION

[0019] The embodiments of the present application realize personalized caries risk assessment and real-time warning through a caries identification, analysis and warning method and system based on tooth images, and provide comprehensive caries management recommendations through historical data analysis and prediction.

[0020] The overall idea of ​​the problem in the embodiment of this application is as follows:

[0021] First, digital imaging technology is used to obtain the user's tooth image, and image processing technology is used to extract key features from the tooth image, such as tooth morphology, color change, and enamel integrity. These features are quantified and converted into feature vector values ​​for comparative analysis with the preset caries feature vector threshold, to preliminarily determine whether the user is at risk of caries, and mark and preliminarily evaluate the risk situation; for users who have been marked as having caries risk, the risk level is further refined, and based on the caries risk level, the system sends the user an alarm or warning of the corresponding level, prompting the user to take timely measures to intervene or seek medical treatment; for users who are not currently at risk of caries, the system does not stop there, and uses recurrent neural network time series analysis technology to deeply mine the user's historical tooth data to predict the possibility of caries risk in the future. Finally, a detailed caries risk detection report is generated, providing users with a one-stop service from risk assessment to prevention recommendations.

[0022] See also Figure 1 The embodiment of the present invention provides a technical solution: a dental caries identification, analysis and early warning method based on a dental image, comprising the following steps: obtaining a user's dental image, and then extracting the user's dental data through the user's dental image; generating a user's dental feature vector value based on the user's dental data, and using the user's dental feature vector value and a dental caries feature vector threshold for analysis to mark and evaluate whether the user has a dental caries risk; further identifying the dental caries risk level for the marked user who has a dental caries risk, and issuing an alarm based on the dental caries risk level; for the user who does not have a dental caries risk, obtaining the user's historical dental data, and using a recurrent neural network to perform a time series analysis on the user's historical dental data to predict the probability of the user having a dental caries risk; and generating a dental caries risk detection report for the user.

[0023] Specifically, the specific analysis of extracting user tooth data through user tooth image is as follows: preprocessing the user tooth image to obtain the preprocessed user tooth image; the preprocessing includes adjusting the image brightness and contrast, image denoising, and image size normalization operations; locating and segmenting the tooth area in the preprocessed user tooth image based on the image segmentation algorithm, and determining the position and boundary of each tooth to obtain the segmented user tooth image; using the feature extraction method to extract the user tooth data from the segmented user tooth image, the user tooth data specifically includes the morphological data of each tooth, the color data of each tooth, and the texture data of each tooth; the morphological data of each tooth is specifically the tooth surface convexity, the color data of each tooth is specifically the tooth surface grayscale value, and the texture data of each tooth is specifically the plaque extension range.

[0024] In this embodiment, the image segmentation algorithm can specifically use methods based on threshold, edge detection, region growth, etc. to determine the boundary and position of each tooth through the segmentation algorithm. Morphological operations and connectivity analysis can usually be used to further extract the area of ​​each tooth. The tooth surface concavity is calculated by detecting the curvature of the tooth edge, the concave and convex features of the surface, or based on the surface morphological analysis. The tooth surface gray value is specifically obtained by converting the color image into a gray image, and then calculating the statistical characteristics of the gray level in the tooth area, such as the average gray value or gray level distribution. The extension range of dental plaque is specifically obtained by using image texture analysis methods, such as local binary patterns or grayscale co-occurrence matrices, to quantify the extension range and density of dental plaque on the tooth surface; the comprehensive use of morphological, color and texture data can improve the recognition accuracy of caries, especially in the recognition of early caries. Through quantitative analysis and comparison of historical data, the risk of caries can be more accurately predicted and real-time alarm notifications can be provided, which helps users and dentists take timely preventive and treatment measures; based on personalized caries risk assessment, targeted oral health management and treatment plans can be formulated to improve treatment effects and user satisfaction.

[0025] By extracting data such as tooth surface convexity and concavity, grayscale value, and plaque extension range, it is possible to detect and analyze subtle changes and abnormalities on the tooth surface, thereby achieving sensitive identification of early caries; based on each user's specific tooth data, a personalized caries risk assessment can be achieved. The tooth structure and characteristics of different users are different, so a personalized assessment can more accurately reflect each person's caries risk; through quantitative morphological, color, and texture data, it is possible to track and analyze the development trend of caries, provide a scientific basis for dentists, and guide the formulation of treatment and preventive measures.

[0026] Specifically, the specific analysis of generating the user's tooth feature vector value based on the user's tooth data is as follows: number each tooth of the user, combine the tooth surface convexity, tooth surface grayscale value and dental plaque extension range of each tooth into the feature vector of each tooth; for the feature vector of each tooth, use support vector machine for training, and output the feature vector value of each tooth.

[0027] In this embodiment, a support vector machine (SVM) is used as a classifier to train the feature vector of each tooth, thereby outputting the feature vector value of each tooth. The support vector machine is a supervised learning algorithm, which is mainly used for binary or multi-classification problems. The specific training process is: the labeled tooth feature vector data set is divided into a training set and a test set, and the feature vectors are standardized or normalized to ensure that the importance of each feature is equal; an appropriate kernel function is selected according to the nature of the data, and common ones include linear kernels, polynomial kernels, and Gaussian kernels; the parameters of the SVM model are optimized through optimization algorithms, such as the sequential minimum optimization algorithm, SMO, so that the model can find the best hyperplane on the training set to segment data points of different categories; the test set is used to evaluate the performance of the trained model, and the accuracy, precision, recall rate, F1 score and other indicators are used to evaluate the effect of the classifier.

[0028] By numerically expressing the specific characteristics of each tooth and using SVM for training, an accurate assessment of caries risk can be achieved. The eigenvector value of each tooth reflects its unique caries risk potential, which helps to detect and prevent dental problems in advance. Based on the specific eigenvector value of each tooth, personalized oral health management suggestions and caries prevention measures can be provided to each user, thereby enhancing treatment effects and user satisfaction. Using machine learning methods such as SVM, potential caries problems can be quickly and accurately identified in large amounts of tooth data, providing dentists with more effective diagnostic support.

[0029] Specifically, the user's tooth feature vector value and the caries feature vector threshold are used for analysis to mark and evaluate whether the user has a caries risk. The specific analysis is as follows: the user's teeth are classified into incisors, canines, premolars and molars based on a deep learning computer vision method, and the caries feature vector thresholds of incisors, canines, premolars and molars are obtained respectively; the feature vector values ​​of each tooth of the user are compared with the caries feature vector thresholds of different classifications according to the user's tooth classification, and teeth with feature vector values ​​greater than or equal to the caries feature vector threshold are marked as teeth with caries risk. When there is a marking situation, the user is assessed to have a caries risk.

[0030] In this implementation scheme, the specific analysis of classifying user teeth into incisors, canines, premolars and molars based on deep learning computer vision method is as follows: obtaining batches of tooth image data, and annotating them with the help of semi-automatic tools in combination with doctor's advice, to ensure that each image has the correct classification label (the classification labels are incisors, canines, premolars, and molars); preprocessing the images, including cropping, scaling, and grayscale, to ensure image quality consistency and suitability for model input, and using data enhancement techniques, such as rotation, flipping, scaling, and brightness adjustment, to increase data diversity and model robustness; selecting deep learning models, such as ResNet, VGG, etc., and selecting classic deep learning models. The neural network structure suitable for the tooth classification task is designed based on the learning architecture. The data set is divided into training set, validation set and test set, usually in a ratio of 70%-15%-15%; the deep learning model is trained on the training set, and the validation set is used for model tuning and hyperparameter adjustment to prevent overfitting; the test set is used to evaluate the accuracy, precision, recall and other indicators of the trained model in classifying tooth types; based on the evaluation results, the model is optimized and adjusted, such as adjusting the network structure, increasing regularization, optimizing the learning rate, etc.; the trained deep learning model is deployed in actual applications to classify the user's teeth and output the classification results of each tooth as incisor, canine, premolar or molar.

[0031] When identifying whether there is caries through the tooth surface convexity, tooth surface gray value and dental plaque extension range of each tooth, the user's teeth need to be classified into incisors, canines, premolars and molars. The reasons are mainly based on the following aspects: incisors are mainly responsible for cutting food, and their tooth surfaces are relatively flat and easy to clean, but they may also cause caries due to food residue and improper cleaning; canines are mainly used to tear food, and their tooth surfaces are sharp and protruding, which may be more likely to accumulate food residues and plaque, thereby increasing the risk of caries; premolars are located between incisors and molars, assisting in cutting food, and their tooth surface morphology and function are between incisors and molars, and their sensitivity to caries is also between the two; molars are located at the innermost part of the mouth, used to grind food, and their tooth surfaces are uneven, with many pits and fissures. These parts are prone to retain plaque and food residues, and are high-incidence areas of caries. The difference in tooth morphology leads to different sensitivities to caries, and the position and arrangement of different teeth in the mouth will also affect the degree of plaque invasion. Therefore, based on the classification of the user's teeth, the caries risk of each tooth can be assessed more accurately, and corresponding prevention and treatment measures can be formulated. At the same time, tooth classification helps doctors to more comprehensively consider various factors during the diagnosis process, such as the morphology, function, position of the teeth, and the patient's oral hygiene habits, thereby improving the accuracy of the diagnosis.

[0032] The caries feature vector is obtained based on the analysis and statistics of a large number of samples. It is used to determine whether a tooth has a potential caries risk. For each tooth, its feature vector value is compared with the caries feature vector threshold of the corresponding classification: for incisors: compare the feature vector value of the incisors with the incisor caries feature vector threshold; for canines: compare the feature vector value of the canines with the canine caries feature vector threshold; for premolars: compare the feature vector value of the premolars with the premolar caries feature vector threshold; for postmolars: compare the feature vector value of the postmolars with the postmolar caries feature vector threshold. By classifying the user's teeth according to type and comparing them based on specific caries feature vector thresholds, a personalized caries risk assessment for each tooth is achieved, which is more accurate and targeted than the general assessment method; when the caries problem has not yet shown obvious symptoms, the potential risk of caries can be discovered in advance through the analysis of the feature vector value, which helps to take preventive measures in time to avoid further deterioration of the disease; and provides scientific basis for dentists to help them formulate personalized treatment plans and oral health management recommendations, thereby improving treatment effects and patient satisfaction.

[0033] The specific method for setting the caries feature vector threshold is as follows: collect a large amount of image data of different types of teeth, and preprocess the images, including brightness adjustment and noise removal; extract features such as concavity, grayscale value and plaque extension range from the image of each tooth; determine the caries feature vector threshold for each type of tooth through statistical analysis or machine learning methods. These thresholds can be set based on the distribution of the data and relevant clinical experience to ensure that normal and risky tooth conditions can be accurately distinguished.

[0034] Specifically, the specific analysis for further identifying the caries risk level of the marked user is as follows: the total number of marks for the marked user is obtained, and then the teeth at each marked position are renumbered; the Euclidean distance calculation is performed on the eigenvector values ​​of the re-numbered teeth and the caries eigenvector thresholds corresponding to different categories, and then the Euclidean distance calculation results of all the re-numbered teeth are averaged to obtain the user's caries risk value, and the user's caries risk level is identified based on the user's caries risk value.

[0035] In this embodiment, the Euclidean distance calculation results of all marked teeth are averaged to obtain the user's caries risk value, which reflects the overall caries risk level of the user's teeth; the caries risk level can be divided into several levels, such as low risk, medium risk and high risk. By calculating the Euclidean distance between the feature vector of each tooth and the threshold, and comprehensively obtaining the overall caries risk value, a quantitative assessment of the user's caries risk is achieved, which is more scientific and accurate; through the alarm mechanism, timely warnings can be issued to users before the caries problem becomes serious, thereby improving the effectiveness of treatment and prevention; the alarm mechanism not only increases users' attention to oral health, but also enables medical staff to respond and deal with caries problems more quickly, thereby improving the efficiency and quality of oral health management; the caries risk level classification standard is based on clinical experience and statistical analysis, and is set according to factors such as the severity and extension range of caries. For example, low risk corresponds to a smaller Euclidean distance threshold, while high risk corresponds to a larger threshold. The following factors are considered in the specific setting: clinical history of teeth, morphological characteristics of teeth; the speed of caries development. By analyzing the data of a large number of patients and the development trend of caries, a caries risk level classification standard that conforms to the actual situation is formulated.

[0036] Specifically, for the situation where the user has no caries risk, the user's historical dental data is obtained, and a recurrent neural network is used to perform time series analysis on the user's historical dental data to predict the probability of the user having caries risk. The specific analysis is as follows: the user's historical dental data includes the tooth surface convexity, tooth surface grayscale value and dental plaque extension range extracted from the user's historically uploaded tooth images; the tooth surface convexity, tooth surface grayscale value and dental plaque extension range extracted from the user's historically uploaded tooth images are used for training to obtain a trained recurrent neural network; the user's latest dental data is obtained, the user's latest dental data is used as the trained recurrent neural network input, the user's latest dental data is predicted, and the prediction result is output, that is, the occurrence of caries at the next time point The probability of tooth risk; the specific training method of the recurrent neural network is: the tooth surface convexity, tooth surface gray value and dental plaque extension range extracted from the tooth images uploaded by the user in history are combined into data sets at each time point in chronological order to form a time series, and then the data sets are divided into training set, validation set and test set; the various hierarchical structures of the recurrent neural network specifically include: the input layer is used to receive historical tooth data, the hidden layer is used to capture the dependencies in the time series, and the output layer is used to predict the probability of caries risk at the next time point; the loss function and optimizer are selected, and then the training set is used for model training, the model parameters are adjusted through the validation set, and the model performance is evaluated using the test set. The model performance includes accuracy, precision, recall rate and F1 score.

[0037] In this implementation scheme, the historical dental data are organized into a time series in chronological order, and each time point corresponds to a set of dental feature data; the historical dental data are divided into a training set, a validation set, and a test set according to the time series, the training set is used for model training, the validation set is used to adjust the model parameters, and the test set is used to finally evaluate the model performance; the recurrent neural network can specifically use RNN or its variants (such as long short-term memory network LSTM or gated recurrent unit GRU) to capture the dependencies in the time series; a loss function suitable for the binary classification problem is selected, such as binary cross entropy, which is used to measure the difference between the predicted value and the actual label. Commonly used optimizers include Adam and SGD, which are used to minimize the loss function; the defined RNN model is trained using the training set, and the model parameters are optimized by the back propagation algorithm; the hyperparameters of the model, such as the learning rate, RNN layer, are adjusted on the validation set. size to optimize the performance and generalization ability of the model; use the test set to evaluate the performance of the trained model. The evaluation indicators may include accuracy, precision, recall, F1 score, etc. to ensure the predictive ability and stability of the model; use the trained RNN model to predict the user's latest dental data and output the prediction result, that is, the probability of the user having caries risk at the next time point; use the RNN model to analyze the time series changes of the user's historical dental data, realize the personalized prediction of the caries risk of each user, and adjust the prediction results according to individual characteristics; when the caries problem has not yet shown obvious symptoms, identify the possible caries risk of the user in advance, and help the user take preventive measures or seek medical treatment in time; the oral medical team can formulate personalized treatment plans based on the prediction results, improve the treatment effect and user satisfaction, and provide users with personalized health management services.

[0038] Specifically, the caries risk detection report includes whether the user is at risk of caries, the mark location where the user is at risk of caries, the user's caries risk level, the probability of the user being at risk of caries, and caries coping suggestions; caries coping suggestions include coping suggestions when the user is at risk of caries and coping suggestions when the user is not at risk of caries; coping suggestions when the user is at risk of caries include medical treatment suggestions, dietary suggestions, and brushing method suggestions; coping suggestions when the user is not at risk of caries include prevention suggestions, dietary suggestions, and brushing method suggestions.

[0039] In this implementation plan, the medical treatment recommendation specifically recommends that users seek medical treatment in a timely manner based on the specific level of dental caries, and be examined and treated by a professional dentist, and provide personalized treatment plans, including fillings, root canal treatments, etc.; regular check-ups are recommended to monitor the treatment effect and progress of dental caries. Dietary recommendations specifically include recommendations to reduce the intake of high-sugar foods and beverages, especially sticky sugars and carbonated beverages; recommend a balanced diet, increase the intake of vegetables and fruits, and provide adequate nutrition to support oral health. Brushing method recommendations specifically include guidance on correct brushing techniques, including brushing twice a day for at least 2 minutes each time, and using dental floss for gap cleaning; it is recommended to use a soft-bristled toothbrush and a suitable toothpaste, and choose a fluoride toothpaste based on the doctor's advice. Prevention recommendations specifically include recommendations for regular oral examinations, usually every six months, to ensure that potential problems are discovered in a timely manner; it is recommended to use fluoride mouthwash to help protect teeth, especially after meals.

[0040] Provide personalized advice based on the user's specific situation and caries risk level. For example, high-risk users require more frequent checkups and stricter diet and brushing instructions. Provide personalized advice and regular reminders to users through personal accounts on the application or website. Specifically, important oral health advice can be sent via push notifications or emails. Follow up on the user's implementation and changes in oral health status to adjust and optimize the advice content to ensure that users continue to improve their oral health.

[0041] By providing preventive advice, it helps users effectively prevent the occurrence of caries and other oral health problems, improves users' oral health awareness, and enhances their understanding of the importance of oral health, thereby improving their overall oral health level, and helps users achieve better oral health management and prevention measures under different caries risk conditions.

[0042] A dental caries recognition, analysis and early warning system based on dental images, which applies the above-mentioned dental caries recognition, analysis and early warning method based on dental images, includes: a data acquisition module, which is used to acquire a user's dental image, and then extract the user's dental data through the user's dental image; a dental caries risk identification module, which is used to generate a user's dental feature vector value based on the user's dental data, and use the user's dental feature vector value and the dental caries feature vector threshold for analysis to mark and evaluate whether the user has a dental caries risk; a risk level identification module, which is used to further identify the dental caries risk level for the marked user with a dental caries risk, and issue an alarm based on the dental caries risk level; a risk prediction module, which is used to acquire the user's historical dental data for the situation where the user does not have a dental caries risk, and use a recurrent neural network to perform time series analysis on the user's historical dental data to predict the probability of the user having a dental caries risk; and a report generation module, which is used to generate a dental caries risk detection report for the user.

[0043] In summary, this application has at least the following effects:

[0044] By acquiring and analyzing the user's dental images and data, the system provides each user with personalized caries risk assessment and warning. The personalized method can more accurately identify the caries risk of each user and provide targeted suggestions and preventive measures. The system instantly analyzes the user's current dental status and conducts real-time caries risk assessment based on the caries feature vector and the preset threshold, helping the user to take necessary preventive measures as early as possible to avoid further deterioration of the caries problem. By utilizing the user's historical dental data and applying technologies such as recurrent neural networks for time series analysis, the system can predict the probability of the user's future caries risk. Predictive analysis helps to formulate personalized prevention plans in advance to reduce the possibility of caries. The caries risk detection report includes the specific location of the user's caries risk, the risk level and the probability of caries risk, providing the user with clear health feedback and response suggestions, which can not only help the user understand the personal oral health status, but also promote the implementation of effective health education and prevention strategies. Early identification and prevention of caries can reduce the medical expenses and patient suffering caused by caries treatment. Early intervention and preventive measures can not only help maintain oral health, but also reduce the patient's economic burden and the inconvenience of the treatment process.

[0045] It should be understood by those skilled in the art that the embodiments of the present invention may be provided as methods and systems. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0046] The present invention is described with reference to the flowcharts and structure diagrams of the methods and systems according to the embodiments of the present invention. It should be understood that each process and combination of modules in the flowcharts and structure diagrams can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts. Figure 1 process or processes and structures Figure 1 A device that specifies functionality in a module or modules.

[0047] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 process or processes and structures Figure 1 A function specified in a module or multiple modules.

[0048] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 process or processes and structures Figure 1 Steps to specify functionality in a module or multiple modules.

[0049] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0050] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A caries recognition, analysis and early warning method based on tooth images, characterized in that: The following steps are involved: Acquire a user's dental image, and then extract the user's dental data through the user's dental image; Generate a user's tooth feature vector value based on the user's tooth data, and use the user's tooth feature vector value and the caries feature vector threshold to perform analysis to mark and evaluate whether the user has a caries risk; Further identifying the caries risk level of the marked user and issuing an alarm based on the caries risk level; For users who do not have the risk of tooth decay, the user's historical tooth data is obtained, and a recurrent neural network is used to perform time series analysis on the user's historical tooth data to predict the probability of the user having the risk of tooth decay; Generate a caries risk detection report for the user.

2. The method for identifying, analyzing and warning dental caries based on tooth images according to claim 1, characterized in that: The specific analysis of extracting user tooth data through user tooth image is as follows: Preprocessing the user's tooth image to obtain a preprocessed user's tooth image; the preprocessing includes adjusting the image brightness and contrast, image denoising, and image size normalization operations; Locate and segment the tooth region in the preprocessed user tooth image based on an image segmentation algorithm, determine the position and boundary of each tooth, and obtain a segmented user tooth image; Extracting user tooth data from the segmented user tooth image using a feature extraction method, wherein the user tooth data specifically includes morphological data of each tooth, color data of each tooth, and texture data of each tooth; The morphological data of each tooth is specifically the tooth surface concavity and convexity, the color data of each tooth is specifically the tooth surface grayscale value, and the texture data of each tooth is specifically the dental plaque extension range.

3. The method for identifying, analyzing and warning dental caries based on tooth images according to claim 1, characterized in that: The specific analysis of generating the user's tooth feature vector value based on the user's tooth data is as follows: Number each tooth of the user, and combine the tooth surface convexity, tooth surface gray value and dental plaque extension range of each tooth into a feature vector of each tooth; The support vector machine is used to train the feature vector of each tooth and output the feature vector value of each tooth.

4. The method for identifying, analyzing and warning dental caries based on tooth images according to claim 1, characterized in that: The specific analysis of using the user's tooth feature vector value and the caries feature vector threshold to mark and evaluate whether the user has caries risk is as follows: A computer vision method based on deep learning classifies the user's teeth into incisors, canines, premolars and molars, and obtains the caries feature vector thresholds of incisors, canines, premolars and molars respectively; According to the user's tooth classification, the feature vector value of each tooth of the user is compared with the caries feature vector threshold of different classifications, and the teeth with feature vector values ​​greater than or equal to the caries feature vector threshold are marked as teeth with caries risk. When there is a marking situation, it is assessed that the user has a caries risk.

5. The method for caries identification, analysis and early warning based on tooth images according to claim 1, characterized in that: The specific analysis for further identifying the caries risk level of the marked user with caries risk is as follows: The total number of marks is obtained for the marked user who has the risk of dental caries, and then the teeth at each marked position are renumbered; The Euclidean distance calculation is performed on the eigenvector values ​​of the secondary numbered teeth and the caries eigenvector thresholds corresponding to different categories, and then the Euclidean distance calculation results of all secondary numbered teeth are averaged to obtain the user's caries risk value, and the user's caries risk value level is identified based on the user's caries risk value.

6. The method for identifying, analyzing and warning dental caries based on tooth images according to claim 1, characterized in that: In the case where the user does not have the risk of dental caries, the user's historical dental data is obtained, and the recurrent neural network is used to perform time series analysis on the user's historical dental data. The specific analysis to predict the probability of the user having the risk of dental caries is as follows: The user's historical tooth data includes tooth surface concavity and convexity, tooth surface gray value and dental plaque extension range extracted from the user's historically uploaded tooth images; The tooth surface concavity, tooth surface gray value and dental plaque extension range extracted from the tooth images uploaded by the user are used for training to obtain a trained recurrent neural network; Obtain the user's latest dental data, use the user's latest dental data as the input of the trained recurrent neural network, predict the user's latest dental data, and output the prediction result, that is, the probability of dental caries risk at the next time point.

7. The method for identifying, analyzing and warning dental caries based on tooth images according to claim 6, characterized in that: The specific training method of the recurrent neural network is: The tooth surface concavity, tooth surface gray value and dental plaque extension range extracted from the tooth images uploaded by the user in history are combined into data sets at each time point in chronological order to form a time series, and then the data sets are divided into training set, validation set and test set; The hierarchical structure of the recurrent neural network specifically includes: the input layer is used to receive historical tooth data, the hidden layer is used to capture the dependencies in the time series, and the output layer is used to predict the probability of caries risk at the next time point; Select the loss function and optimizer, then use the training set to train the model, adjust the model parameters through the validation set, and use the test set to evaluate the model performance. The model performance includes accuracy, precision, recall, and F1 score.

8. The method for caries identification, analysis and early warning based on tooth images according to claim 1, characterized in that: The caries risk detection report specifically includes whether the user has caries risk, the marked location of the user's caries risk, the user's caries risk level, the probability of the user having caries risk, and caries coping suggestions; The dental caries coping suggestions specifically include coping suggestions when the user is at risk of dental caries and coping suggestions when the user is not at risk of dental caries; When users are at risk of tooth decay, the response suggestions include medical treatment suggestions, dietary suggestions, and tooth brushing method suggestions; The response suggestions for users who are not at risk of tooth decay include prevention suggestions, dietary suggestions, and brushing method suggestions.

9. A caries recognition, analysis and early warning system based on tooth images, using the caries recognition, analysis and early warning method based on tooth images as claimed in any one of claims 1 to 8, characterized in that: include: A data acquisition module, used to acquire a user's tooth image, and then extract the user's tooth data through the user's tooth image; A dental caries risk identification module is used to generate a user's dental feature vector value based on the user's dental data, and use the user's dental feature vector value and the dental caries feature vector threshold to perform analysis to mark and evaluate whether the user has a dental caries risk; A risk level identification module is used to further identify the caries risk level of the marked user with caries risk and issue an alarm based on the caries risk level; The risk prediction module is used to obtain the user's historical dental data for the situation that the user does not have the risk of dental caries, and use the recurrent neural network to perform time series analysis on the user's historical dental data to predict the probability of the user having the risk of dental caries; The report generation module is used to generate a caries risk detection report for the user.

Citation Information

Cited By

  • Dental diagnosis and treatment scheme generation method and system based on user image data

    CN120496801A

  • A method and system for generating dental treatment plans based on user image data

    CN120496801B

  • Dental caries early warning system based on image recognition

    CN121260464A

  • Dental caries identification method and system based on deep learning

    CN121505399A