Oral anterior tooth anti-jaw detection method and system based on image recognition

Abstract

The invention discloses an oral anterior tooth anti-jaw detection method and system based on image recognition. The method comprises the following steps: step 1, image acquisition: acquiring an image of an oral anterior tooth area of a patient by using a special oral imaging device; step 2, image preprocessing: preprocessing the collected image; step 3, tooth region segmentation: carrying out tooth region segmentation on the preprocessed image by using a semantic segmentation algorithm based on deep learning; 4, feature extraction: extracting features related to anterior tooth reverse jaw from the segmented tooth region image; step 5, performing anti-jaw judgment: according to the extracted features, combining a preset anti-jaw judgment rule; and step 6, outputting a result: outputting the detection result in a visual mode. The semantic segmentation algorithm based on deep learning is adopted for tooth region segmentation, so that the teeth can be accurately separated from other tissues in the oral cavity, the interference of other tissues on tooth feature extraction is reduced, and the detection accuracy is greatly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of oral anterior teeth crossbite detection, and in particular to an oral anterior teeth crossbite detection method and system based on image recognition. Background Art

[0002] Anterior crossbite is a common oral malocclusion, commonly known as "underbite". This deformity not only affects the patient's facial appearance, but also has adverse effects on oral functions such as chewing and pronunciation. Long-term existence may even cause complications such as temporomandibular joint disease. Therefore, early and accurate detection of anterior crossbite is of great significance for formulating a reasonable treatment plan and improving the patient's oral health and quality of life.

[0003] Traditional methods for detecting anterior crossbite in the oral cavity mainly rely on clinical examinations and manual measurements by dentists. Doctors observe the arrangement of teeth in the patient's mouth, the occlusion relationship, and use dental models for measurement and analysis. However, this method has many limitations. First, its test results largely depend on the doctor's experience and professional level, and the judgment criteria between different doctors may vary, resulting in strong subjectivity in the test results. Secondly, the manual measurement process is cumbersome and inefficient, and it is difficult to meet the needs of large-scale oral disease screening. In addition, for some mild cases of anterior crossbite, it may be easy to miss the diagnosis by relying solely on naked eye observation and manual measurement.

[0004] With the rapid development of computer technology and image processing technology, image recognition technology is increasingly used in the medical field. In stomatology, the use of image recognition technology to analyze oral X-rays, oral photos, etc. can assist doctors in diagnosing diseases more accurately and efficiently. However, the current oral anterior teeth crossbite detection technology based on image recognition is still in the development stage, and there are some problems that need to be solved. For example, when processing complex oral images, the existing image recognition algorithm is not accurate enough in extracting tooth features and is easily interfered by other tissues in the oral cavity, resulting in low accuracy of the detection results. At the same time, some detection systems lack adaptability to images under different shooting conditions. When the image has problems such as uneven lighting and shooting angle deviation, the detection effect will be greatly affected.

[0005] In summary, it is of great practical significance to develop a method and system with high accuracy, strong adaptability and the ability to quickly detect oral anterior teeth crossbite. Therefore, the present invention proposes an oral anterior teeth crossbite detection method and system based on image recognition. Summary of the invention

[0006] The object of the present invention is to solve at least one of the technical problems existing in the prior art, and to provide an anterior crossbite detection method and system based on image recognition. By using a semantic segmentation algorithm based on deep learning for tooth region segmentation, the teeth can be accurately separated from other tissues in the oral cavity, reducing the interference of other tissues on tooth feature extraction. At the same time, through multi-dimensional feature extraction and comprehensive judgment rules, the anterior crossbite situation can be judged more comprehensively and accurately. Compared with traditional detection methods, the detection accuracy is greatly improved.

[0007] The image preprocessing module in the system performs processing such as denoising and contrast enhancement on images under different shooting conditions, which can effectively improve the image quality and make the system have better adaptability to images under different lighting conditions and shooting angles. Even when there is a certain degree of uneven illumination or shooting angle deviation in the image, the anterior crossbite can still be accurately detected.

[0008] The model update module of the system can regularly collect new clinical data and update and train the semantic segmentation model and anterior crossbite judgment model, enabling the system to continuously adapt to new cases and features and maintain high detection accuracy and reliability. As the clinical data continues to accumulate, the performance of the system will continue to improve. Through performance comparison tests before and after model update, it is found that the updated model has a significant improvement in detection accuracy.

[0009] The present invention also provides an anterior crossbite detection method based on image recognition as described above, including the following steps:

[0010] Step 1, image acquisition: Use a dedicated oral imaging device to acquire images of the anterior tooth region of the patient, including frontal images and lateral images inside the oral cavity, ensuring that the images clearly and completely show the morphology, position, and occlusion relationship of the anterior teeth;

[0011] Step 2, image preprocessing: Preprocess the acquired images, including image denoising, grayscale conversion, and contrast enhancement operations;

[0012] Step 3, tooth region segmentation: Use a semantic segmentation algorithm based on deep learning to segment the tooth region of the preprocessed image, separate the teeth from other tissues in the oral cavity, construct a U-Net model, and use a large number of labeled oral image data to train the model so that it can accurately identify and segment the tooth region;

[0013] Step 4, feature extraction: Extract features related to anterior crossbite from the segmented tooth region images, including the position features, morphological features, and occlusion features of the teeth;

[0014] Step 5, Crossbite Judgment: Based on the extracted features and in combination with the preset crossbite judgment rules, determine whether the patient has anterior crossbite. If the overjet of the upper and lower anterior teeth is less than the preset threshold and the position of the lower anterior teeth is relatively too far forward compared to the upper anterior teeth, it is determined as anterior crossbite.

[0015] Step 6, Result Output: Output the detection results in a visual manner, including generating a crossbite detection report, which shows the patient's basic information, image acquisition information, detection results, whether there is crossbite, the severity of crossbite, and corresponding suggestions.

[0016] According to an oral anterior crossbite detection method based on image recognition provided by the present invention, for the image preprocessing, noise interference in the image is removed through a denoising algorithm to improve the clarity of the image. Grayscale processing converts the color image into a grayscale image to facilitate subsequent feature extraction and analysis. The histogram equalization method is used for contrast enhancement to highlight the edge and detail features of the teeth.

[0017] According to an oral anterior crossbite detection method based on image recognition provided by the present invention, for the tooth region segmentation, the cross-entropy loss function is used to measure the difference between the model prediction result and the true label, and the stochastic gradient descent optimization algorithm is used to update the model parameters to improve the segmentation accuracy of the model.

[0018] According to an oral anterior crossbite detection method based on image recognition provided by the present invention, in the feature extraction, the position feature is represented by calculating the centroid coordinates of each anterior tooth and the distance parameters between adjacent teeth. The morphological features extract the contour shape, crown length, and root length information of the teeth. The occlusion feature is obtained by analyzing the relative position relationship and overjet of the upper and lower anterior teeth.

[0019] According to an oral anterior crossbite detection method based on image recognition provided by the present invention, for the calculation of the position feature in the feature extraction, using the coordinate system of the image, the centroid coordinates and distance parameters are obtained by statistically calculating the pixel points in the tooth region. For the morphological features, the contour extraction algorithm is used to extract the contour of the teeth, and then the length parameters of the crown and root are calculated. For the occlusion feature, by comparing the position relationship of the upper and lower anterior teeth in the image, the overjet index is calculated.

[0020] According to an oral anterior crossbite detection method based on image recognition provided by the present invention, the crossbite judgment classifies it into mild, moderate, and severe crossbites according to the severity of the crossbite. Mild crossbite is manifested as the lower anterior teeth slightly covering the upper anterior teeth, with little impact on oral function and aesthetics. Moderate crossbite shows that the lower anterior teeth cover the upper anterior teeth to a more obvious degree, and oral function may be affected to a certain extent. Severe crossbite means that the lower anterior teeth severely cover the upper anterior teeth, causing a greater impact on oral function and facial aesthetics.

[0021] According to an anterior crossbite detection method based on image recognition provided by the present invention, the result output marks the positions and features of the crossbite teeth on the image, so that doctors can more intuitively understand the patient's condition. When generating a detection report, a templatized method is adopted to automatically fill in relevant information according to the detection results, improving the efficiency and accuracy of report generation.

[0022] According to an anterior crossbite detection system based on image recognition provided by the present invention, it includes:

[0023] Image acquisition module: used to acquire images of the anterior teeth area of the patient's oral cavity, including frontal and lateral images inside the oral cavity, and record relevant information during image acquisition. This module consists of a dedicated oral imaging device and a data acquisition terminal, and the data acquisition terminal is used to store and transmit the acquired images and relevant information;

[0024] Image preprocessing module: adopts image processing algorithms, including Gaussian filtering, median filtering, and histogram equalization, and automatically selects a suitable algorithm for processing according to the characteristics of the image. When automatically selecting an algorithm, the corresponding denoising algorithm is selected by analyzing the type and intensity of the noise in the image;

[0025] Tooth area segmentation module: based on the semantic segmentation algorithm of deep learning, accurately segment the tooth area from the oral image; use the U-Net model and train the model with a large amount of labeled data to enable it to have the ability to accurately identify and segment the tooth area;

[0026] Feature extraction module: extract the position, shape, and occlusion features related to anterior crossbite from the segmented tooth area image, providing a basis for crossbite judgment. Use a method based on geometric calculation to obtain the position features, a method based on contour analysis to extract the shape features, and a method based on position relationship comparison to obtain the occlusion features;

[0027] Crossbite judgment module: judge whether the patient has anterior crossbite according to the extracted features and the preset crossbite judgment rules, and determine the severity of the crossbite;

[0028] Result output module: present the detection results in a visual way, generate an anterior crossbite detection report, and mark crossbite-related information on the image; it includes a report generation sub-module and an image annotation sub-module. The report generation sub-module automatically generates a detection report according to the detection results, and the image annotation sub-module uses an image annotation tool to mark the positions and features of the crossbite teeth on the image.

[0029] Compared with the prior art, a method and system for detecting anterior crossbite of the oral cavity based on image recognition according to the present invention uses a semantic segmentation algorithm based on deep learning to segment the tooth region, which can accurately separate the teeth from other tissues in the oral cavity and reduce the interference of other tissues on the extraction of tooth features. At the same time, through multi-dimensional feature extraction and comprehensive judgment rules, the anterior crossbite situation can be judged more comprehensively and accurately. Compared with traditional detection methods, the accuracy of detection is greatly improved.

[0030] Compared with the prior art, a method and system for detecting anterior crossbite of the oral cavity based on image recognition according to the present invention. The image preprocessing module in the system performs denoising, contrast enhancement, etc. on images under different shooting conditions, which can effectively improve the image quality and make the system have better adaptability to images under different lighting conditions and shooting angles. Even when there is a certain degree of uneven illumination or shooting angle deviation in the image, the anterior crossbite can still be accurately detected.

[0031] Compared with the prior art, a method and system for detecting anterior crossbite of the oral cavity based on image recognition according to the present invention. The model update module of the system can regularly collect new clinical data and update and train the semantic segmentation model and the anterior crossbite judgment model, so that the system can continuously adapt to new cases and features and maintain high detection accuracy and reliability. With the continuous accumulation of clinical data, the performance of the system will be continuously improved. Through the performance comparison test before and after the model update, it is found that the updated model has a significant improvement in detection accuracy. Description of the Drawings

[0032] The present invention will be further described below in conjunction with the drawings and embodiments;

[0033] Figure 1 It is a flowchart of a method for detecting anterior crossbite of the oral cavity based on image recognition according to the present invention;

[0034] Figure 2 It is an architecture diagram of a system for detecting anterior crossbite of the oral cavity based on image recognition according to the present invention. Detailed Embodiments

[0035] This part will describe the specific embodiments of the present invention in detail. The preferred embodiments of the present invention are shown in the drawings. The function of the drawings is to supplement the description of the text part of the specification, enabling people to intuitively and vividly understand each technical feature and the overall technical solution of the present invention, but it cannot be understood as a limitation on the protection scope of the present invention.

[0036] Refer to Figure 1-2 , an embodiment of a method and system for detecting anterior crossbite of the oral cavity based on image recognition according to the present invention includes:

[0037] Image acquisition: Use a high-definition endoscope device dedicated to the oral cavity to acquire images of the anterior teeth area of the patient's oral cavity. Before acquisition, inform the patient to keep the oral cavity clean to avoid food residues affecting the image quality. Adjust the angle and focal length of the endoscope to ensure that the frontal and lateral images of the oral cavity can clearly show the morphology, position, and occlusion relationship of the anterior teeth. At the same time, through the built-in recording function of the device, record information such as device parameters, shooting angle, and lighting conditions during image acquisition. After acquisition, transfer the image data to the data acquisition terminal for storage and subsequent processing.

[0038] During the acquisition process, if the image quality assessment sub-module detects that the image clarity is lower than the set threshold, there are some teeth not fully displayed, or the lighting unevenness is too high, the system automatically prompts the patient to re-acquire the image until the image quality meets the requirements.

[0039] Image preprocessing: Denoise the acquired images and select an appropriate denoising algorithm according to the type of image noise. If the image mainly contains Gaussian noise, use the Gaussian filtering algorithm for denoising. Gaussian filtering smooths the image by weighted averaging each pixel point in the image and its neighboring pixel points to reduce noise interference. Set the kernel size of the Gaussian filter to [3×3] and the standard deviation to [1.5], and filter the image. If there is salt-and-pepper noise in the image, use the median filtering algorithm, which sorts the pixel values in the neighborhood and takes the median value as the output value of the current pixel, and can effectively remove salt-and-pepper noise. Grayscale processing: Convert the denoised color image to a grayscale image using the weighted average method. The calculation formula is: Gray = 0.299×R + 0.587×G + 0.114×B

[0040] Where R, G, and B represent the pixel values of the red, green, and blue channels in the color image, respectively, and Gray represents the converted grayscale value. Through grayscale processing, the image is converted from the R, G, B color space to the grayscale space, facilitating subsequent feature extraction and analysis.

[0041] Contrast Enhancement: The histogram equalization method is used to enhance the contrast of grayscale images. Histogram equalization redistributes the grayscale values of the image to make the grayscale distribution of the image more uniform, thereby enhancing the contrast of the image. The specific implementation process is as follows: First, count the frequency of each grayscale value in the image to obtain the grayscale histogram; then calculate the cumulative distribution function according to the grayscale histogram; finally, map the grayscale value of each pixel point in the image according to the cumulative distribution function to obtain the image with enhanced contrast. After the contrast enhancement process, the edges and detail features of the teeth are more obvious, which is beneficial to subsequent tooth region segmentation and feature extraction. Tooth Region Segmentation A tooth region segmentation network based on the U-Net model is constructed. The U-Net model is a classic semantic segmentation model with an encoder and a decoder structure, which can effectively extract the features of the image and perform accurate segmentation. When training the U-Net model, a large number of labeled oral image data are collected as the training set.

[0042] The acquisition method of the labeled data is as follows: Professional dentists manually label the tooth regions in the oral images and mark the contours of each tooth.

[0043] The training set images are divided into training data and validation data according to a certain ratio, with 80% for training and 20% for validation. During the training process, the preprocessed images are input into the U-Net model. The model extracts the features of the image through operations such as convolutional layers and pooling layers, and then performs upsampling through transposed convolutional layers to restore the resolution of the image. Finally, the segmented tooth region image is output. The cross-entropy loss function is used to measure the difference between the model prediction result and the true label, and the Adam optimization algorithm is used to update the model parameters. The learning rate is set to [0.001]. During the training process, the model is regularly evaluated using the validation data, and indicators such as the accuracy and recall rate of the model are observed. When the performance of the model on the validation set no longer improves, the training is stopped to obtain the trained tooth region segmentation model. Feature Extraction Location Feature Extraction: For the segmented tooth region image, calculate the centroid coordinates of each anterior tooth.

[0044] By traversing the pixel points of the tooth region, according to the centroid calculation formula: and where x i , y i are the coordinates of the pixel points in the tooth region, and n is the total number of pixel points in the tooth region, the centroid coordinates of each anterior tooth are obtained. At the same time, calculate the distance between adjacent teeth, which is represented by calculating the Euclidean distance between the centroid coordinates of adjacent teeth. For two adjacent teeth A and B, their centroid coordinates are (x A , y A ) and (x B , y B ).

[0045] The embodiments of the present invention have been described in detail above in conjunction with the accompanying drawings. However, the present invention is not limited to the above embodiments, and various changes can be made without departing from the spirit of the present invention within the scope of knowledge possessed by those of ordinary skill in the art.

Claims

1. An anterior crossbite detection method for the oral cavity based on image recognition, characterized in that, It includes the following steps: Step 1, Image acquisition: Use a dedicated oral imaging device to acquire images of the anterior teeth area of the patient's oral cavity, including frontal and lateral images inside the oral cavity, ensuring that the images clearly and completely show the morphology, position, and occlusion relationship of the anterior teeth; Step 2, Image preprocessing: Preprocess the acquired images, including image denoising, grayscale conversion, and contrast enhancement operations; Step 3, Tooth area segmentation: Use a semantic segmentation algorithm based on deep learning to segment the tooth area from the preprocessed images, separating the teeth from other tissues inside the oral cavity. Construct a U-Net model and use a large number of labeled oral image data to train the model so that it can accurately identify and segment the tooth area; Step 4, Feature extraction: Extract features related to anterior crossbite from the segmented tooth area images, including the position features, morphological features, and occlusion features of the teeth; Step 5, Crossbite judgment: According to the extracted features and combined with the preset crossbite judgment rules, judge whether the patient has anterior crossbite. If the overjet degree of the upper and lower anterior teeth is less than the preset threshold and the position of the lower anterior teeth is too far forward relative to the upper anterior teeth, it is determined as anterior crossbite; Step 6, Result output: Output the detection results in a visual way, including generating a crossbite detection report, which shows the patient's basic information, image acquisition information, and detection results, as well as whether there is crossbite, the severity of crossbite, and corresponding suggestions.

2. The method for detecting anterior crossbite of the oral cavity based on image recognition according to claim 1, wherein, For the image preprocessing, the noise interference in the image is removed through a denoising algorithm to improve the clarity of the image. The grayscale conversion process converts the color image into a grayscale image, which is convenient for subsequent feature extraction and analysis. The histogram equalization method is used for contrast enhancement to highlight the edge and detail features of the teeth.

3. The method for detecting anterior crossbite of the oral cavity based on image recognition according to claim 1, wherein For the tooth area segmentation, the cross-entropy loss function is used to measure the difference between the model prediction result and the true label, and the stochastic gradient descent optimization algorithm is used to update the model parameters to improve the segmentation accuracy of the model.

4. The method for detecting anterior crossbite of the oral cavity based on image recognition according to claim 1, characterized in that, In the feature extraction, the position features are represented by calculating the centroid coordinates of each anterior tooth and the distance parameters between adjacent teeth. The morphological features extract the contour shape, crown length, and root length information of the teeth. The occlusion features are obtained by analyzing the relative position relationship and overjet degree of the upper and lower anterior teeth.

5. The method for detecting anterior crossbite of the oral cavity based on image recognition according to claim 1, wherein In the feature extraction, for the calculation of the position features, using the coordinate system of the image, the centroid coordinates and distance parameters are obtained by statistically calculating the pixel points in the tooth area. For the morphological features, a contour extraction algorithm is used to extract the contour of the teeth, and then the length parameters of the crown and root are calculated. For the occlusion features, by comparing the position relationship of the upper and lower anterior teeth in the image, the overjet degree index is calculated.

6. The method for detecting anterior crossbite of the oral cavity based on image recognition according to claim 1, wherein According to the severity of crossbite, the crossbite judgment classifies it into mild, moderate, and severe crossbite. Mild crossbite is manifested as the lower anterior teeth slightly covering the upper anterior teeth, with little impact on oral function and aesthetics. Moderate crossbite shows a more obvious degree of the lower anterior teeth covering the upper anterior teeth, and oral function may be affected to a certain extent. Severe crossbite means that the lower anterior teeth severely cover the upper anterior teeth, causing a greater impact on oral function and facial aesthetics.

7. A method for detecting anterior crossbite of the oral cavity based on image recognition according to claim 1, wherein, The result output marks the positions and features of the cross-bite teeth on the image, so that doctors can more intuitively understand the patient's condition. When generating the detection report, a templated method is adopted to automatically fill in relevant information according to the detection results, improving the efficiency and accuracy of report generation.

8. An anterior crossbite detection system for the oral cavity based on image recognition, which adopts an anterior crossbite detection method for the oral cavity based on image recognition as described in any one of claims 1-7, characterized in that, It includes: Image acquisition module: used to acquire images of the anterior teeth area of the patient's oral cavity, including frontal and lateral images inside the oral cavity, and record relevant information during image acquisition. This module consists of a dedicated oral imaging device and a data acquisition terminal, and the data acquisition terminal is used to store and transmit the acquired images and relevant information; Image preprocessing module: adopts image processing algorithms, including Gaussian filtering, median filtering, and histogram equalization, and automatically selects a suitable algorithm for processing according to the characteristics of the image. When automatically selecting an algorithm, the corresponding denoising algorithm is selected by analyzing the noise type and intensity of the image; Tooth area segmentation module: based on the semantic segmentation algorithm of deep learning, accurately segments the tooth area from the oral cavity image; uses the U-Net model and trains the model with a large amount of labeled data to enable it to accurately identify and segment the tooth area; Feature extraction module: extracts the position, morphology, and occlusion features related to anterior cross-bite from the segmented tooth area image, providing a basis for cross-bite judgment. Geometric calculation-based methods are used to obtain position features, contour analysis-based methods are used to extract morphological features, and position relationship comparison-based methods are used to obtain occlusion features; Cross-bite judgment module: judges whether the patient has anterior cross-bite according to the extracted features and the preset cross-bite judgment rules, and determines the severity of the cross-bite; Result output module: presents the detection results in a visual way, generates a cross-bite detection report, and marks cross-bite-related information on the image; It includes a report generation sub-module and an image annotation sub-module. The report generation sub-module automatically generates a detection report according to the detection results, and the image annotation sub-module uses an image annotation tool to mark the positions and features of the cross-bite teeth on the image.

Images