Intelligent analysis method and system for oral cavity image

Abstract

The invention discloses an intelligent analysis method and system of an oral cavity image, relates to the technical field of image processing, and aims to calculate a tooth deviation index YPZ through intelligent processing and analysis of the oral cavity image and accurately reflect the overall deviation degree of tooth arrangement and occlusion relationship. Through comparison with a preset clinical statistical deviation threshold LZ, an oral suggestion evaluation scheme can be generated, the deviation index and the deviation threshold are iteratively optimized according to feedback information of a patient, in the traditional orthodontic process, a doctor depends on experience and manual judgment, subjective errors may exist, and the adjustment process is lack of a real-time feedback mechanism, so that the accuracy of orthodontic treatment is poor. The lagging of the treatment effect is caused. Through a feedback and adjustment mechanism of the tooth deviation index YPZ and the clinical statistical deviation threshold LZ, dynamic evaluation of the oral health condition and real-time optimization of the treatment scheme are realized by using a back propagation algorithm, and the accuracy and efficiency of the treatment effect are greatly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and in particular to an intelligent analysis method and system for oral images. Background Art

[0002] With the rapid development of modern medical technology, image analysis, as an interdisciplinary field, has become one of the core technologies of many medical applications. Its application ranges from medical imaging to surgical assistance, covering the entire process from disease diagnosis to treatment planning. With the continuous progress of oral medicine, intelligent analysis of oral images has gradually become one of the important tools in oral medicine, especially in orthodontics, restoration and treatment plan formulation, the analysis of oral images can provide more accurate information.

[0003] At present, in the process of evaluating tooth arrangement and occlusion, most dentists rely on traditional manual operations and experience to make judgments, usually by visual inspection or using traditional X-ray and CT images to observe the position and occlusion of teeth. Although these methods can provide certain reference information, they have significant disadvantages. First, relying on manual visual inspection or experience for evaluation is easily affected by the doctor's personal subjective judgment, which may lead to inconsistent diagnostic results. Secondly, traditional imaging technology has certain limitations in image resolution, contrast and clarity, especially in the slight changes of teeth and complex occlusion relationships, it is difficult to provide accurate information. Moreover, doctors often need to spend a lot of time on manual measurement and comparison, which is time-consuming and prone to errors. Therefore, the lack of automated, high-precision and efficient evaluation tools has seriously restricted further development. Summary of the invention

[0004] In view of the deficiencies of the prior art, the present invention provides an intelligent analysis method and system for oral images, which solves the problems mentioned in the background technology.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: an intelligent analysis method for oral images, comprising the following steps:

[0006] S1, performing denoising preprocessing, contrast enhancement preprocessing and sharpening preprocessing on the collected oral cavity image to obtain a processed oral cavity image;

[0007] S2, using edge detection algorithm and image segmentation method to extract the contour and position of the teeth from the processed oral image, obtain the coordinate information Pt and contour data Ct of each tooth, and form a tooth position feature set Ft;

[0008] S3, comparing the tooth position feature set Ft with the preset calibration model to obtain the deviation value of each tooth to form a tooth deviation vector Dt;

[0009] S4, performing statistical analysis based on the tooth deviation vector Dt to obtain a tooth deviation index YPZ to reflect the overall deviation degree of the tooth arrangement and occlusal relationship;

[0010] S5. Obtain the image oral suggestion evaluation plan by comparing the tooth deviation index YPZ with the preset clinical statistical deviation threshold LZ, collect feedback information based on the content of the image oral suggestion evaluation plan, and then use the back propagation mechanism to iteratively adjust the tooth deviation index YPZ and the clinical statistical deviation threshold LZ.

[0011] Preferably, said S1 includes S11 and S12;

[0012] S11, performing denoising preprocessing on the collected oral image, including using Gaussian filtering to remove noise in the oral image, and obtaining a filtered oral image;

[0013] S12, using histogram equalization to perform contrast enhancement preprocessing on the filtered oral image, and then using the Laplace operator to sharpen the oral image to highlight edge information in the oral image, thereby obtaining a sharpened oral image.

[0014] Preferably, S2 includes S21 and S22;

[0015] S21. Use an edge detection algorithm to identify the edges of teeth in the oral image after sharpening, including using the Canny edge detection algorithm to identify the edges of teeth in the oral image, specifically including smoothing, gradient calculation, non-maximum suppression and double threshold detection processing steps, and finally output a binary image to mark the edges in the image, separate the teeth and the background in the oral image, and obtain the tooth edge image Ee.

[0016] Preferably, S22, using an image segmentation method to extract the contour and position of the teeth from the edge image Ee, including extracting the specific area of ​​each tooth, and then calculating the precise coordinate position of each tooth, using a threshold-based region growing algorithm or other segmentation techniques for segmentation, segmenting the edge image Ee by using a threshold segmentation method, and when the pixel value of the edge image is greater than the threshold, the pixel is considered to be part of the tooth, and it is judged whether each pixel belongs to the tooth area, and the specific area of ​​the tooth is marked by the white area in the edge image Ee, and the segmented tooth image S is obtained, and then the contour tracking and boundary detection algorithm is used to extract the edge coordinates of each tooth from the tooth image S to generate contour data Ct, and then the contour data Ct is geometrically calculated to obtain the center coordinates of each tooth, and the coordinate information Pt is generated to form a tooth position feature set Ft;

[0017] The coordinate information Pt specifically represents the geometric center coordinates of teeth 1 to n, and the coordinate information Pt={(x1, y1), (x2, y3), ..., (xn, yn)}, (xn, yn) represents the geometric center coordinates (x, y) of the nth tooth;

[0018] The contour data Ct specifically includes the two-dimensional coordinates of the contours of teeth 1 to n, and the contour data Ct={{(x11, y11), (x12, y12), ...)}, {(x21, yi1), (x22, y22), ...)}, ..., {(xn1, yn1), (xn2, yn2), ..., (xnk, ynk)}}; (x11, y11) represents the first contour coordinate position (x, y) of the first tooth, (x12, y12) represents the second contour coordinate position (x, y) of the first tooth, and (xnk, ynk) represents the kth contour coordinate position (x, y) of the nth tooth;

[0019] The position feature set Ft specifically includes contour data Ct and coordinate information Pt of 1 to n teeth, and the position feature set Ft is specifically Ft={(Ct, n, Pt, n)|i=1, 2, ..., n}.

[0020] Preferably, said S3 includes S31;

[0021] S31. Substitute the coordinate information Pt of each tooth in the tooth position feature set Ft into the preset calibration model and use the Euclidean distance method to calculate, obtain the position deviation ratio △Pt of each tooth, and integrate them simultaneously to form a tooth deviation vector Dt, wherein the tooth deviation vector Dt specifically includes the position deviation ratio △Pt of teeth 1 to n, and the tooth deviation vector Dt = {△Pt1, △Pt2, ..., △Ptn}. The preset calibration model is based on the ideal position of teeth under normal occlusion and arrangement.

[0022] Preferably, said S4 includes S41 and S42;

[0023] S41, by performing statistical calculation on the position deviation ratio △Pt of each tooth in the tooth deviation vector Dt, the average value of the tooth deviation is obtained, which is marked as the tooth deviation average index △Pj, and then the degree of dispersion of the position deviation ratio △Pt of each tooth is analyzed, which is marked as the tooth deviation standard deviation △Pc;

[0024] The tooth deviation average index ΔPj is obtained by the following calculation formula:

[0025]

[0026]

[0027] Wherein, n represents teeth from 1 to n, specifically the total number of teeth, i represents the i-th tooth, xi and yi represent the horizontal and vertical positions of the geometric center coordinates of the i-th tooth, respectively, MDx and MDy represent the horizontal mean deviation and vertical mean deviation, respectively, and XZ and YZ represent the horizontal mean and vertical mean in the calibration model, respectively;

[0028] The tooth deviation standard deviation ΔPc is obtained by the following calculation formula:

[0029]

[0030] Where MCx and MCy represent the standard deviation of the horizontal axis and the standard deviation of the vertical axis, respectively.

[0031] Preferably, S42, weighted calculation is performed according to the obtained tooth deviation average index △Pj and tooth deviation standard deviation △Pc to obtain the tooth deviation index YPZ, and quantify the tooth arrangement and occlusion relationship, including when the YPZ value is large, it indicates that the patient's tooth deviation is serious, and there may be obvious irregular arrangement or occlusion problems; when the YPZ value is small, it indicates that the patient's teeth are arranged more neatly, the occlusion relationship is more normal, and reflects the overall deviation degree of the tooth arrangement and occlusion relationship;

[0032] The tooth deviation index YPZ is obtained by the following calculation formula:

[0033] YPZ=z1*ΔPj+z2*ΔPc;

[0034] Wherein, z1 and z2 represent the preset weight values ​​of the mean index △Pj of tooth deviation and the standard deviation △Pc of tooth deviation respectively, and z1+z2=1, and the specific value is set by the user.

[0035] Preferably, said S5 includes S51 and S52;

[0036] S51, by comparing the tooth deviation index YPZ with the preset clinical statistical deviation threshold LZ, obtaining an image oral suggestion evaluation scheme, wherein the image oral suggestion evaluation scheme is obtained by the following comparison method:

[0037] When the tooth deviation index YPZ ≤ the clinical statistical deviation threshold LZ, obtain the image oral cavity evaluation plan with normal tooth arrangement and occlusion;

[0038] When the tooth deviation index YPZ>clinical statistical deviation threshold LZ, obtain the oral image suggestion evaluation plan for tooth arrangement and occlusion abnormality, prompt the patient to adjust the observation frequency, and mark the oral image source patient evaluation mark as BJ=2;

[0039] When the tooth deviation index YPZ>clinical statistical deviation threshold LZ*150%, obtain the oral image suggestion evaluation plan for tooth arrangement and occlusion abnormality, prompt the patient to undergo orthodontic treatment, and mark the oral image source patient evaluation mark as BJ=5;

[0040] When the tooth deviation index YPZ>clinical statistical deviation threshold LZ*200%, obtain the image oral recommendation evaluation plan for tooth arrangement and occlusion warning, prompt the patient to perform tooth arrangement correction and tooth occlusion adjustment, and mark the oral image source patient evaluation mark as oral evaluation mark BJ=9.

[0041] Preferably, S52, feedback information is collected according to the content of the image oral recommendation evaluation plan, including feedback information collected from patient follow-up records and feedback information collected from mini-program questionnaires, wherein the time period for collecting feedback information is scaled by the oral evaluation mark BJ, and the feedback information is subjected to steps S1 to S4 to obtain the feedback index FZ, and then the error index E is calculated with the tooth deviation index YPZ, and the gradient descent method of the back-propagation mechanism is used to iteratively adjust the tooth deviation index YPZ and the clinical statistical deviation threshold LZ, including scaling the error index E with the tooth deviation index YPZ and the clinical statistical deviation threshold LZ.

[0042] An intelligent analysis system for oral images, comprising an image processing module, a feature extraction module, a deviation calculation module, an evaluation module and a feedback iteration module;

[0043] The image processing module performs denoising preprocessing, contrast enhancement preprocessing and sharpening preprocessing on the collected oral cavity image to obtain a processed oral cavity image;

[0044] The feature extraction module uses edge detection algorithm and image segmentation method to extract the contour and position of the teeth from the processed oral image, obtains the coordinate information Pt and contour data Ct of each tooth, and forms a tooth position feature set Ft;

[0045] The deviation calculation module compares the tooth position feature set Ft with the preset calibration model to obtain the deviation value of each tooth to form a tooth deviation vector Dt;

[0046] The evaluation module performs statistical analysis based on the tooth deviation vector Dt to obtain the tooth deviation index YPZ, which reflects the overall deviation degree of the tooth arrangement and occlusal relationship;

[0047] The feedback iteration module compares the tooth deviation index YPZ with the preset clinical statistical deviation threshold LZ to obtain the image oral suggestion evaluation plan, collects feedback information according to the content of the image oral suggestion evaluation plan, and then uses the back propagation mechanism to iteratively adjust the tooth deviation index YPZ and the clinical statistical deviation threshold LZ.

[0048] The present invention provides an intelligent analysis method and system for oral images, which have the following beneficial effects:

[0049] (1) Through intelligent processing and analysis of oral images, the tooth deviation index YPZ is calculated to accurately reflect the overall deviation degree of tooth arrangement and occlusal relationship. By comparing with the preset clinical statistical deviation threshold LZ, an oral recommendation evaluation plan can be generated, and the deviation index and deviation threshold are iteratively optimized according to patient feedback information. In the traditional orthodontic process, doctors rely on experience and manual judgment, which may result in subjective errors, and the adjustment process lacks a real-time feedback mechanism, resulting in a lag in treatment effect. Through the feedback and adjustment mechanism of the tooth deviation index YPZ and the clinical statistical deviation threshold LZ, the back propagation algorithm is used to realize the dynamic evaluation of oral health status and real-time optimization of treatment plans, greatly improving the accuracy and efficiency of treatment effects.

[0050] (2) By using the Euclidean distance method to substitute the coordinate information Pt of each tooth in the tooth position feature set Ft into the preset calibration model for comparison, the position deviation ratio △Pt of each tooth is accurately calculated, and these deviation data are integrated into the tooth deviation vector Dt. This process can accurately reflect the deviation of each tooth relative to the ideal arrangement position, providing a solid data foundation for subsequent deviation statistical analysis. Then, by statistically analyzing the position deviation ratio △Pt of each tooth in the tooth deviation vector Dt, the tooth deviation average index △Pj and the tooth deviation standard deviation △Pc are generated, reflecting the overall deviation degree of tooth arrangement and occlusion relationship. Through weighted calculation, the tooth deviation index YPZ is further obtained, which quantifies the tooth arrangement and occlusion relationship, facilitating clinical evaluation and diagnosis.

[0051] (3) By comparing the tooth deviation index YPZ with the preset clinical statistical deviation threshold LZ, the patient's tooth arrangement and occlusion status can be accurately identified, flexible and personalized oral health advice can be provided, and the patient's assessment mark BJ can be adjusted in time to make the assessment results more accurate and dynamic. By collecting feedback information such as patient follow-up records and mini-program questionnaires, the understanding of the patient's oral health status is further deepened, and the feedback data is processed using the feedback index FZ. The error index E is calculated by comparing with the tooth deviation index YPZ, and the tooth deviation index YPZ and the clinical statistical deviation threshold LZ are continuously adjusted, and the gradient descent method of the back-propagation mechanism is used for iterative optimization. This process ensures the continuous precision of the oral health assessment plan, can reflect the changes in the patient's oral health in real time, and ensures that each treatment or prevention recommendation can be optimized and adjusted as the patient's status changes. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 A schematic diagram of the steps of an intelligent analysis method for oral images of the present invention;

[0053] Figure 2 The figure is a schematic diagram of a block diagram of an intelligent analysis system for oral images according to the present invention. DETAILED DESCRIPTION

[0054] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0055] Example 1

[0056] The present invention provides an intelligent analysis method for oral images. Figure 1 , including the following steps:

[0057] S1, performing denoising preprocessing, contrast enhancement preprocessing and sharpening preprocessing on the collected oral cavity image to obtain a processed oral cavity image;

[0058] S2, using edge detection algorithm and image segmentation method to extract the contour and position of the teeth from the processed oral image, obtain the coordinate information Pt and contour data Ct of each tooth, and form a tooth position feature set Ft;

[0059] S3, comparing the tooth position feature set Ft with the preset calibration model to obtain the deviation value of each tooth to form a tooth deviation vector Dt;

[0060] S4, performing statistical analysis based on the tooth deviation vector Dt to obtain a tooth deviation index YPZ to reflect the overall deviation degree of the tooth arrangement and occlusal relationship;

[0061] S5. Obtain the image oral suggestion evaluation plan by comparing the tooth deviation index YPZ with the preset clinical statistical deviation threshold LZ, collect feedback information based on the content of the image oral suggestion evaluation plan, and then use the back propagation mechanism to iteratively adjust the tooth deviation index YPZ and the clinical statistical deviation threshold LZ.

[0062] In this embodiment, through the intelligent processing and analysis of oral images, an efficient, accurate and personalized tooth correction scheme evaluation tool is provided, which can automatically evaluate the arrangement and occlusion of teeth through advanced image processing and machine learning technology without changing the hardware. Specifically, by building a tooth position feature set Ft and comparing it with the calibration model, the system can quickly obtain the deviation value of each tooth, and further calculate the tooth deviation index YPZ, which accurately reflects the overall deviation degree of tooth arrangement and occlusion. By comparing with the preset clinical statistical deviation threshold LZ, the method can generate an oral suggestion evaluation scheme, and iteratively optimize the deviation index and deviation threshold according to patient feedback information to ensure the accuracy and personalization of the scheme. In the traditional tooth correction process, doctors rely on experience and manual judgment, which may have subjective errors, and the adjustment process lacks a real-time feedback mechanism, resulting in a lag in the treatment effect. The present invention uses the feedback and adjustment mechanism of the tooth deviation index YPZ and the clinical statistical deviation threshold LZ, and uses the back propagation algorithm to realize the dynamic evaluation of oral health status and the real-time optimization of the treatment scheme, which greatly improves the accuracy and efficiency of the treatment effect. In addition, the system's combined application of image processing and edge detection algorithms not only reduces the need for human intervention, but also significantly reduces the occurrence of human errors and improves the accuracy of treatment plans.

[0063] Example 2

[0064] This embodiment is explained in Example 1, please refer to Figure 1 , specifically: the S1 includes S11 and S12;

[0065] S11, performing denoising preprocessing on the collected oral image, including using Gaussian filtering to remove noise in the oral image, and obtaining a filtered oral image;

[0066] S12, using histogram equalization to perform contrast enhancement preprocessing on the filtered oral image, and then using the Laplace operator to sharpen the oral image to highlight edge information in the oral image, thereby obtaining a sharpened oral image.

[0067] The S2 includes S21 and S22;

[0068] S21. Use an edge detection algorithm to identify the edges of teeth in the oral image after sharpening, including using the Canny edge detection algorithm to identify the edges of teeth in the oral image, specifically including smoothing, gradient calculation, non-maximum suppression and double threshold detection processing steps, and finally output a binary image to mark the edges in the image, separate the teeth and the background in the oral image, and obtain the tooth edge image Ee.

[0069] S22. Use image segmentation methods to extract the contour and position of the teeth from the edge image Ee, including extracting the specific area of ​​each tooth, and then calculating the precise coordinate position of each tooth, using a threshold-based region growing algorithm or other segmentation techniques for segmentation, and segmenting the edge image Ee by using a threshold segmentation method. When the pixel value of the edge image is greater than the threshold, the pixel is considered to be part of the tooth, and it is judged whether each pixel belongs to the tooth area, and the white area in the edge image Ee is used to mark the specific area of ​​the tooth, and the segmented tooth image S is obtained, and then the contour tracking and boundary detection algorithm is used to extract the edge coordinates of each tooth from the tooth image S to generate contour data Ct, and then the contour data Ct is geometrically calculated to obtain the center coordinates of each tooth, and the coordinate information Pt is generated to form a tooth position feature set Ft;

[0070] The coordinate information Pt specifically represents the geometric center coordinates of teeth 1 to n, and the coordinate information Pt={(x1, y1), (x2, y3), ..., (xn, yn)}, (xn, yn) represents the geometric center coordinates (x, y) of the nth tooth;

[0071] The contour data Ct specifically includes the two-dimensional coordinates of the contours of teeth 1 to n, and the contour data Ct={{(x11, y11), (x12, y12), ...)}, {(x21, yi1), (x22, y22), ...)}, ..., {(xn1, yn1), (xn2, yn2), ..., (xnk, ynk)}}; (x11, y11) represents the first contour coordinate position (x, y) of the first tooth, (x12, y12) represents the second contour coordinate position (x, y) of the first tooth, and (xnk, ynk) represents the kth contour coordinate position (x, y) of the nth tooth;

[0072] The position feature set Ft specifically includes contour data Ct and coordinate information Pt of 1 to n teeth, and the position feature set Ft is specifically Ft={(Ct, n, Pt, n)|i=1, 2, ..., n}.

[0073] In this embodiment, the complete process from image acquisition to accurate tooth identification combines multiple advanced technologies such as denoising, contrast enhancement, sharpening, edge detection and image segmentation to accurately extract the tooth position feature set Ft in the oral image. First, through denoising preprocessing and contrast enhancement, the quality of the oral image is guaranteed, the noise is removed and the details are enhanced, so that the subsequent edge detection and image segmentation can be more accurate. Then, the edge detection algorithm and image segmentation method use the Canny edge detection algorithm to accurately identify the edge of the tooth, and further extract the contour and precise position of the tooth through the threshold segmentation technology, and generate the tooth position feature set Ft, which includes the precise geometric center coordinates Pt and two-dimensional contour coordinates Ct of each tooth. These processing steps work together to ensure the high-precision acquisition of tooth position information, providing a solid data foundation for subsequent deviation analysis and personalized treatment plans. Through this process, the present invention realizes high-precision tooth positioning and segmentation in oral image analysis, avoiding the problems of misidentification or incomplete processing caused by poor image quality or inaccurate algorithms in traditional methods. The effective generation of the tooth position feature set Ft ensures that the subsequent analysis of tooth deviation can be based on accurate geometric information, greatly improving the application effect of the system in the field of orthodontics. This method can overcome the instability and errors caused by traditional manual labeling, improve the efficiency and accuracy of tooth position and morphology analysis, avoid the limitations of human operation, have strong adaptability and high stability in image processing for different patients, and ensure accurate identification and deviation calculation of each tooth.

[0074] Example 3

[0075] This embodiment is explained in Example 2. Please refer to Figure 1 , specifically: the S3 includes S31;

[0076] S31. Substitute the coordinate information Pt of each tooth in the tooth position feature set Ft into the preset calibration model and use the Euclidean distance method to calculate, obtain the position deviation ratio △Pt of each tooth, and integrate them simultaneously to form a tooth deviation vector Dt, wherein the tooth deviation vector Dt specifically includes the position deviation ratio △Pt of teeth 1 to n, and the tooth deviation vector Dt = {△Pt1, △Pt2, ..., △Ptn}. The preset calibration model is based on the ideal position of teeth under normal occlusion and arrangement.

[0077] The calibration model includes a preset ideal tooth arrangement and occlusal position feature set Ft. Specifically, an ideal position is obtained by combining multiple normal people's tooth data sets and obtaining an average ideal tooth arrangement data set through statistical methods.

[0078] The S4 includes S41 and S42;

[0079] S41, by performing statistical calculation on the position deviation ratio △Pt of each tooth in the tooth deviation vector Dt, the average value of the tooth deviation is obtained, which is marked as the tooth deviation average index △Pj, and then the degree of dispersion of the position deviation ratio △Pt of each tooth is analyzed, which is marked as the tooth deviation standard deviation △Pc;

[0080] The tooth deviation average index ΔPj is obtained by the following calculation formula:

[0081]

[0082] Where n represents teeth from 1 to n, specifically the total number of teeth, i represents the i-th tooth, xi and yi represent the horizontal and vertical positions of the geometric center coordinates of the i-th tooth, MDx and MDy represent the horizontal mean deviation and vertical mean deviation, respectively, and XZ and YZ represent the horizontal mean and vertical mean in the calibration model, respectively;

[0083] The tooth deviation standard deviation ΔPc is obtained by the following calculation formula:

[0084]

[0085] Where MCx and MCy represent the standard deviation of the horizontal axis and the standard deviation of the vertical axis, respectively.

[0086] S42, performing weighted calculation based on the obtained tooth deviation average index △Pj and tooth deviation standard deviation △Pc to obtain the tooth deviation index YPZ, and quantifying the tooth arrangement and occlusion relationship, including when the YPZ value is large, it means that the patient's tooth deviation is serious, and there may be obvious irregular arrangement or occlusion problems; when the YPZ value is small, it means that the patient's teeth are arranged more neatly and the occlusion relationship is more normal, and reflects the overall deviation degree of the tooth arrangement and occlusion relationship;

[0087] The tooth deviation index YPZ is obtained by the following calculation formula:

[0088] YPZ=z1*ΔPj+z2*ΔPc;

[0089] Wherein, z1 and z2 represent the preset weight values ​​of the mean index △Pj of tooth deviation and the standard deviation △Pc of tooth deviation respectively, and z1+z2=1, and the specific value is set by the user.

[0090] In this embodiment, the coordinate information Pt of each tooth in the tooth position feature set Ft is substituted into the preset calibration model for comparison by using the Euclidean distance method, and the position deviation ratio △Pt of each tooth is accurately calculated, and these deviation data are integrated into the tooth deviation vector Dt. This process can accurately reflect the deviation of each tooth relative to the ideal arrangement position, and provide a solid data basis for the subsequent deviation statistical analysis. Then, by statistically analyzing the position deviation ratio △Pt of each tooth in the tooth deviation vector Dt, the tooth deviation average index △Pj and the tooth deviation standard deviation △Pc are generated, reflecting the overall deviation degree of the tooth arrangement and occlusion relationship. Through weighted calculation, the tooth deviation index YPZ is further obtained, which quantifies the tooth arrangement and occlusion relationship, which is convenient for clinical evaluation and diagnosis, not only improving the accuracy of oral image analysis, but also providing an objective and quantifiable evaluation method through the quantitative calculation of the tooth deviation index YPZ, so that tooth arrangement and occlusion problems can be diagnosed quickly and accurately. Different from the traditional manual judgment method, the present invention avoids the interference of human errors and ensures the accurate reflection of the degree of tooth deviation by combining the average index of tooth deviation △Pj and the standard deviation of tooth deviation △Pc. By accurately quantifying YPZ, the degree of abnormality of tooth arrangement and occlusion can be clearly pointed out, providing a scientific basis for subsequent treatment plans, greatly improving the accuracy of clinical decision-making and treatment effects. In addition, the use of a weighted calculation method makes the evaluation of the tooth deviation index YPZ more flexible, and the weight parameters can be adjusted according to specific clinical needs, further improving the applicability and accuracy of the method.

[0091] Example 4

[0092] This embodiment is explained in Example 3, please refer to Figure 1 , specifically: the S5 includes S51 and S52;

[0093] S51, by comparing the tooth deviation index YPZ with the preset clinical statistical deviation threshold LZ, obtaining an image oral suggestion evaluation scheme, wherein the image oral suggestion evaluation scheme is obtained by the following comparison method:

[0094] When the tooth deviation index YPZ ≤ the clinical statistical deviation threshold LZ, obtain the image oral cavity evaluation plan with normal tooth arrangement and occlusion;

[0095] When the tooth deviation index YPZ>clinical statistical deviation threshold LZ, obtain the oral image suggestion evaluation plan for tooth arrangement and occlusion abnormality, prompt the patient to adjust the observation frequency, and mark the oral image source patient evaluation mark as BJ=2;

[0096] When the tooth deviation index YPZ>clinical statistical deviation threshold LZ*150%, obtain the oral image suggestion evaluation plan for tooth arrangement and occlusion abnormality, prompt the patient to undergo orthodontic treatment, and mark the oral image source patient evaluation mark as BJ=5;

[0097] When the tooth deviation index YPZ>clinical statistical deviation threshold LZ*200%, obtain the image oral recommendation evaluation plan for tooth arrangement and occlusion warning, prompt the patient to perform tooth arrangement correction and tooth occlusion adjustment, and mark the oral image source patient evaluation mark as oral evaluation mark BJ=9.

[0098] S52. Collect feedback information according to the content of the oral image suggestion evaluation plan, including collecting feedback information from patient follow-up records and collecting feedback information from the mini-program questionnaire, wherein the time period for collecting feedback information is scaled by the oral evaluation mark BJ, and the feedback information is subjected to steps S1 to S4 to obtain the feedback index FZ, which is then combined with the tooth deviation index YPZ to calculate the error index E, and the gradient descent method of the back-propagation mechanism is used to iteratively adjust the tooth deviation index YPZ and the clinical statistical deviation threshold LZ, including scaling the error index E to the tooth deviation index YPZ and the clinical statistical deviation threshold LZ;

[0099] The error index E is specifically calculated by comparing the feedback index FZ with the previously calculated tooth deviation index YPZ. The error index E represents the difference between the feedback information and the current deviation, and is a measure of the deviation adjustment process. At this time, the system uses the back propagation mechanism and the gradient descent method to further adjust the tooth deviation index YPZ and the clinical statistical deviation threshold LZ.

[0100] In this embodiment, by comparing the tooth deviation index YPZ with the preset clinical statistical deviation threshold LZ, the patient's tooth arrangement and occlusion status can be accurately identified. By setting multiple critical values, when the tooth deviation index YPZ exceeds different critical thresholds, the evaluation plan can be gradually adjusted, from prompting slight abnormalities to suggesting tooth correction or warning of serious problems. This multi-level evaluation mechanism can provide flexible and personalized oral health advice when different degrees of deviation occur, and timely adjust the patient's evaluation mark BJ, so that the evaluation results are more accurate and dynamic. More importantly, by collecting feedback information such as patient follow-up records and mini-program questionnaires, the understanding of the patient's oral health status is further deepened, and the feedback data is processed using the feedback index FZ. By bringing these feedback information into steps S1 to S4 for reprocessing, and comparing with the tooth deviation index YPZ to calculate the error index E, the tooth deviation index YPZ and the clinical statistical deviation threshold LZ are continuously adjusted, and the gradient descent method of the back-propagation mechanism is used for iterative optimization. This process ensures that the oral health assessment plan is continuously refined, reflecting changes in the patient's oral health in real time and ensuring that each treatment or prevention recommendation can be optimized and adjusted as the patient's condition changes.

[0101] Example 5

[0102] An intelligent analysis system for oral images, please refer to Figure 2 ,Specifically: it includes image processing module, feature extraction module, deviation calculation module, evaluation module and feedback iteration module;

[0103] The image processing module performs denoising preprocessing, contrast enhancement preprocessing and sharpening preprocessing on the collected oral cavity image to obtain a processed oral cavity image;

[0104] The feature extraction module uses edge detection algorithm and image segmentation method to extract the contour and position of the teeth from the processed oral image, obtains the coordinate information Pt and contour data Ct of each tooth, and forms a tooth position feature set Ft;

[0105] The deviation calculation module compares the tooth position feature set Ft with the preset calibration model to obtain the deviation value of each tooth to form a tooth deviation vector Dt;

[0106] The evaluation module performs statistical analysis based on the tooth deviation vector Dt to obtain the tooth deviation index YPZ, which reflects the overall deviation degree of the tooth arrangement and occlusal relationship;

[0107] The feedback iteration module compares the tooth deviation index YPZ with the preset clinical statistical deviation threshold LZ to obtain the image oral suggestion evaluation plan, collects feedback information according to the content of the image oral suggestion evaluation plan, and then uses the back propagation mechanism to iteratively adjust the tooth deviation index YPZ and the clinical statistical deviation threshold LZ.

[0108] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An intelligent analysis method for oral images, characterized in that: The following steps are involved: S1, performing denoising preprocessing, contrast enhancement preprocessing and sharpening preprocessing on the collected oral cavity image to obtain a processed oral cavity image; S2, using edge detection algorithm and image segmentation method to extract the contour and position of the teeth from the processed oral image, obtain the coordinate information Pt and contour data Ct of each tooth, and form a tooth position feature set Ft; S3, comparing the tooth position feature set Ft with the preset calibration model to obtain the deviation value of each tooth to form a tooth deviation vector Dt; S4, performing statistical analysis based on the tooth deviation vector Dt to obtain a tooth deviation index YPZ to reflect the overall deviation degree of the tooth arrangement and occlusal relationship; S5. Obtain the image oral suggestion evaluation plan by comparing the tooth deviation index YPZ with the preset clinical statistical deviation threshold LZ, collect feedback information based on the content of the image oral suggestion evaluation plan, and then use the back propagation mechanism to iteratively adjust the tooth deviation index YPZ and the clinical statistical deviation threshold LZ.

2. The intelligent analysis method of oral images according to claim 1, characterized in that: Said S1 includes S11 and S12; S11, performing denoising preprocessing on the collected oral image, including using Gaussian filtering to remove noise in the oral image, and obtaining a filtered oral image; S12, using histogram equalization to perform contrast enhancement preprocessing on the filtered oral image, and then using the Laplace operator to sharpen the oral image to highlight edge information in the oral image, thereby obtaining a sharpened oral image.

3. The intelligent analysis method of oral images according to claim 2, characterized in that: The S2 includes S21 and S22; S21. Using an edge detection algorithm to identify the edges of teeth in the oral image after the sharpening process, including using a Canny edge detection algorithm to identify the edges of teeth in the oral image, separating the teeth from the background in the oral image, and obtaining a tooth edge image Ee.

4. The intelligent analysis method of oral images according to claim 3, characterized in that: S22. Use an image segmentation method to extract the contour and position of the teeth from the edge image Ee, including using a threshold segmentation method to segment the edge image Ee, using the white area in the edge image Ee to mark the specific area of ​​the teeth, obtaining the segmented tooth image S, and then using a contour tracking and boundary detection algorithm to extract the edge coordinates of each tooth from the tooth image S to generate contour data Ct, and then performing geometric calculations on the contour data Ct to obtain the center coordinates of each tooth, generate coordinate information Pt, and form a tooth position feature set Ft.

5. The intelligent analysis method of oral images according to claim 4, characterized in that: The S3 includes S31; S31. Substitute the coordinate information Pt of each tooth in the tooth position feature set Ft into the preset calibration model and use the Euclidean distance method to calculate to obtain the position deviation ratio △Pt of each tooth, and integrate them synchronously to form a tooth deviation vector Dt.

6. The intelligent analysis method of oral images according to claim 5, characterized in that: The S4 includes S41 and S42; S41. By performing statistical calculation on the position deviation ratio △Pt of each tooth in the tooth deviation vector Dt, the average value of the tooth deviation is obtained, which is marked as the tooth deviation average index △Pj. Then, the degree of discreteness of the position deviation ratio △Pt of each tooth is analyzed, which is marked as the tooth deviation standard deviation △Pc.

7. The intelligent analysis method of oral images according to claim 6, characterized in that: S42. Perform weighted calculation based on the obtained tooth deviation average index △Pj and tooth deviation standard deviation △Pc to obtain the tooth deviation index YPZ, quantify the relationship between tooth arrangement and occlusion, and reflect the overall deviation degree of the relationship between tooth arrangement and occlusion.

8. The intelligent analysis method of oral images according to claim 1, characterized in that: The S5 includes S51 and S52; S51, by comparing the tooth deviation index YPZ with the preset clinical statistical deviation threshold LZ, obtaining an image oral suggestion evaluation scheme, wherein the image oral suggestion evaluation scheme is obtained by the following comparison method: When the tooth deviation index YPZ ≤ the clinical statistical deviation threshold LZ, obtain the image oral cavity evaluation plan with normal tooth arrangement and occlusion; When the tooth deviation index YPZ>clinical statistical deviation threshold LZ, obtain the oral image suggestion evaluation plan for tooth arrangement and occlusion abnormality, prompt the patient to adjust the observation frequency, and mark the oral image source patient evaluation mark as BJ=2; When the tooth deviation index YPZ>clinical statistical deviation threshold LZ*150%, obtain the oral image suggestion evaluation plan for tooth arrangement and occlusion abnormality, prompt the patient to undergo orthodontic treatment, and mark the oral image source patient evaluation mark as BJ=5; When the tooth deviation index YPZ>clinical statistical deviation threshold LZ*200%, obtain the image oral recommendation evaluation plan for tooth arrangement and occlusion warning, prompt the patient to perform tooth arrangement correction and tooth occlusion adjustment, and mark the oral image source patient evaluation mark as oral evaluation mark BJ=9.

9. The intelligent analysis method of oral images according to claim 1, characterized in that: S52. Collect feedback information based on the content of the image oral evaluation suggestion plan, including collecting feedback information from patient follow-up records and collecting feedback information from the mini-program questionnaire, wherein the time period for collecting feedback information is scaled by the oral evaluation mark BJ, and the feedback information is subjected to steps S1 to S4 to obtain the feedback index FZ, which is then combined with the tooth deviation index YPZ to calculate the error index E, and the gradient descent method of the back-propagation mechanism is used to iteratively adjust the tooth deviation index YPZ and the clinical statistical deviation threshold LZ, including scaling the error index E to the tooth deviation index YPZ and the clinical statistical deviation threshold LZ.

10. An intelligent analysis system for oral images, applied to an intelligent analysis method for oral images according to any one of claims 1 to 9, characterized in that: It includes image processing module, feature extraction module, deviation calculation module, evaluation module and feedback iteration module; The image processing module performs denoising preprocessing, contrast enhancement preprocessing and sharpening preprocessing on the collected oral cavity image to obtain a processed oral cavity image; The feature extraction module uses edge detection algorithm and image segmentation method to extract the contour and position of the teeth from the processed oral image, obtains the coordinate information Pt and contour data Ct of each tooth, and forms a tooth position feature set Ft; The deviation calculation module compares the tooth position feature set Ft with the preset calibration model to obtain the deviation value of each tooth to form a tooth deviation vector Dt; The evaluation module performs statistical analysis based on the tooth deviation vector Dt to obtain the tooth deviation index YPZ, which reflects the overall deviation degree of the tooth arrangement and occlusal relationship; The feedback iteration module compares the tooth deviation index YPZ with the preset clinical statistical deviation threshold LZ to obtain the image oral suggestion evaluation plan, collects feedback information according to the content of the image oral suggestion evaluation plan, and then uses the back propagation mechanism to iteratively adjust the tooth deviation index YPZ and the clinical statistical deviation threshold LZ.

Images