Method for measuring buccal-lingual torque value of maxillary molar based on improved ResNET

Abstract

The embodiment of the invention discloses a method for measuring a buccal-lingual torque value of maxillary molar based on improved ResNET. The method comprises the following steps: acquiring and preprocessing three-dimensional images of teeth and alveolar bones, and performing three-dimensional reconstruction on the preprocessed images; based on the coupling relationship between the dental crown and the tooth root, positioning a first tooth long axis; positioning a second tooth long axis based on the coupling relationship between the tooth and the alveolar bone; calculating an included angle between the first tooth long axis and the second tooth long axis and generating a new long axis; calculating and verifying a buccal-lingual torque value of the maxillary molar; through the automatic process, the accurate measurement and calculation of the upper jaw first molar torque value are realized, and the effectiveness and accuracy are improved.

Description

Technical Field

[0001] The embodiments of the present application relate to the technical field of torque value measurement, and in particular to a method for measuring the buccolingual torque value of maxillary molars based on improved ResNET. Background Art

[0002] With the rapid development of modern oral medicine and artificial intelligence technology, the accurate measurement of tooth posture values has become increasingly important in orthodontic treatment. The posture value is an important parameter describing the spatial angle of the tooth long axis relative to a reference plane (such as the maxillary plane PP), and its accurate measurement directly affects the formulation of the design of orthodontic appliances. Through the accurate evaluation of the posture value, doctors can design more scientific orthodontic treatment plans. However, there are still certain limitations in the measurement efficiency and accuracy of the existing technologies. Summary of the Invention

[0003] The embodiments of the present application provide a method for measuring the buccolingual torque value of maxillary molars based on improved ResNET, which solves the problems of limited feature extraction, high data requirements, large computational complexity, and insufficient model generalization ability of the existing methods.

[0004] In a first aspect, the embodiments of the present application provide a method for measuring the buccolingual torque value of maxillary molars based on improved ResNET, and the method includes the following steps: Obtain three-dimensional images of teeth and alveolar bone and perform preprocessing, and perform three-dimensional reconstruction on the preprocessed images; Locate the first tooth long axis based on the coupling relationship between the crown and the root; Locate the second tooth long axis based on the coupling relationship between the tooth and the alveolar bone; Calculate the angle between the first tooth long axis and the second tooth long axis and generate a new long axis; Calculate the buccolingual torque value of the maxillary molar and verify it.

[0005] Further, the obtaining three-dimensional images of teeth and alveolar bone and performing preprocessing, and performing three-dimensional reconstruction on the preprocessed images includes: Use a cone beam computed tomography (CBCT) device to obtain three-dimensional images of teeth and alveolar bone; For the three-dimensional images, apply median filtering or Gaussian filtering technology to remove the noise introduced during scanning, use contrast-limited adaptive histogram equalization (CLAHE) technology to enhance the contrast between teeth and alveolar bone, and extract the region of interest (ROI); Use open-source tools to perform three-dimensional reconstruction on the preprocessed images to generate a visual model of teeth and alveolar bone; wherein, the three-dimensional reconstruction includes image segmentation, feature extraction, and three-dimensional modeling.

[0006] Further, the positioning of the first tooth long axis based on the coupling relationship between the dental crown and the tooth root includes: Using region growing or threshold segmentation algorithms to separate the dental crown and the tooth root; starting from a seed point through the region growing algorithm, gradually adding neighboring pixels similar to the seed point until the entire tooth region is covered, and through threshold segmentation, the tooth is distinguished from the background based on the range of pixel values, usually requiring prior histogram analysis to determine the optimal threshold; Using U-Net or its variant network to automatically segment the tooth region; Extracting point cloud data from the segmented dental crown and tooth root regions, where the point cloud data includes the three-dimensional coordinate information of the tooth; Applying principal component analysis (PCA) to calculate the principal axis directions of the dental crown and the tooth root. PCA projects the data into a new coordinate system through linear transformation, such that the variance on the principal component of the first coordinate axis is the largest to determine the principal axis direction; Using the least squares method to fit the midlines of the dental crown and the tooth root; among them, the best fit line is found by minimizing the sum of the squares of the errors, and the midlines of the dental crown and the tooth root are connected to form the first tooth long axis; Displaying the fitting result through a visualization interface; If it is found that the long axis deviates severely, adjust the position of the connection point between the dental crown and the tooth root.

[0007] Further, the positioning of the second tooth long axis based on the coupling relationship between the tooth and the alveolar bone includes: Performing three-dimensional curvature analysis on the alveolar bone region to identify the main boundaries of the alveolar bone by calculating the curvature of the alveolar bone surface; Selecting the intersection point of the extension lines such that the focus is located inside the alveolar bone; Using the direction vectors of the extension lines to calculate their bisecting direction, and the formula is: ; Among them, vector L1 and vector L2 are the direction vectors of two extension lines passing through the contact points between the tooth and the alveolar bone; the direction vectors are bisected to determine a direction that takes into account both the tooth and the alveolar bone; the bisector is used as the second tooth long axis, and its spatial coordinates are recorded.

[0008] Further, the calculation of the included angle between the first tooth long axis and the second tooth long axis and the generation of a new long axis include: Starting from the direction vectors of the two tooth long axes, where the vectors are obtained from the long axis data obtained by fitting and define the direction of the long axis in three-dimensional space; Using the dot product formula to calculate the cosine value of the included angle between the two vectors, and the dot product formula is: ; Among them, a and b are the two vectors, and is the modulus of the vector, and θ is the included angle between the vectors; The included angle θ is obtained through the arccosine function. This included angle represents the spatial relationship between the two long axes and is an important parameter for evaluating tooth arrangement and alveolar bone structure; A new long axis is generated using the angle bisector. The direction vector of the angle bisector can be obtained through the average value of the normalized sum of the two original long axis direction vectors, that is: ; The new long axis is recorded and visualized to match the actual anatomical structure; during the visualization process, the position and direction of the new long axis are adjusted.

[0009] Furthermore, calculating and verifying the buccolingual torque value of the maxillary molar includes: Determining the included angle between the new long axis and the alveolar bone plane, and calculating the included angle through the dot product of vectors; among them, the dot product of the direction vector of the new long axis and the normal vector of the alveolar bone plane is used to obtain the result for calculating the cosine value of the included angle, and then the included angle is obtained; Calculating the horizontal and vertical offsets of the crown top and root bottom along the new long axis; Determining the exact positions of the crown and root, and measuring the offsets relative to the new long axis; The calculation formula for the buccolingual torque value of the maxillary molar is expressed as: ; Among them, θ is the included angle between the new long axis and the alveolar bone plane, and Δy and Δx are the horizontal and vertical offsets of the crown top and root bottom along the new long axis respectively; Comparing the calculated torque value with the normal anatomical reference value or the clinical orthodontic target value; If the measurement error is large, check the quality of the CBCT image or the accuracy of the segmentation model. Among them, the image quality check includes the evaluation of parameters such as resolution, noise level, and contrast, and the accuracy check of the segmentation model is to retrain or adjust the model parameters.

[0010] In a second aspect, the embodiments of the present application further provide a measuring device for the buccolingual torque value of the maxillary molar based on the improved ResNET, including: An image acquisition module for acquiring three-dimensional images of teeth and alveolar bone, performing preprocessing, and performing three-dimensional reconstruction on the preprocessed images; A first positioning module for positioning the first tooth long axis based on the coupling relationship between the crown and the root; A second positioning module for positioning the second tooth long axis based on the coupling relationship between the tooth and the alveolar bone; An included angle calculation module, configured to calculate the included angle between the first tooth long axis and the second tooth long axis and generate a new long axis; A torque calculation module, configured to calculate the buccolingual torque value of the maxillary molar and perform verification.

[0011] In a third aspect, an embodiment of the present application further provides a computer device, including: a memory and one or more processors; The memory is configured to store one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement a method for measuring the buccolingual torque value of the maxillary molar based on the improved ResNET as described above.

[0012] In a fourth aspect, an embodiment of the present application further provides a storage medium including computer-executable instructions, and the computer-executable instructions are used to execute a method for measuring the buccolingual torque value of the maxillary molar based on the improved ResNET as described above when executed by a computer processor.

[0013] In a fifth aspect, an embodiment of the present application provides a computer program product, and the computer program product includes instructions that, when executed by a computer, cause the computer to implement the method as described above.

[0014] The embodiment of the present application acquires three-dimensional images of teeth and alveolar bone and performs preprocessing, and performs three-dimensional reconstruction on the preprocessed images; based on the coupling relationship between the tooth crown and the tooth root, locates the first tooth long axis; based on the coupling relationship between the tooth and the alveolar bone, locates the second tooth long axis; calculates the included angle between the first tooth long axis and the second tooth long axis and generates a new long axis; calculates the buccolingual torque value of the maxillary molar and performs verification; through an automated process, realizes the accurate measurement and calculation of the torque value, and improves the effectiveness and accuracy. Description of the Drawings

[0015] Figure 1 is a flowchart of a method for measuring the buccolingual torque value of the maxillary molar based on the improved ResNET provided by an embodiment of the present application; Figure 2 is a structural schematic diagram of a device for measuring the buccolingual torque value of the maxillary molar based on the improved ResNET provided by an embodiment of the present application; Figure 3 is a structural schematic diagram of a computer device provided by an embodiment of the present application; Figure 4 is a schematic diagram of torque value measurement of a method for measuring the buccolingual torque value of the maxillary molar based on the improved ResNET provided by an embodiment of the present application; Figure 5It is a schematic diagram of the calibration baseline of a method for measuring the buccolingual torque value of maxillary molars based on the improved ResNET provided by an embodiment of the present application; Figure 6 It is a schematic diagram of the MLP neural network structure of a method for measuring the buccolingual torque value of maxillary molars based on the improved ResNET provided by an embodiment of the present application. Detailed implementation manners

[0016] In order to make the objectives, technical solutions and advantages of the present application clearer, the following further describes the specific embodiments of the present application in detail with reference to the accompanying drawings. It can be understood that the specific embodiments described herein are only used to explain the present application, rather than limiting the present application. Additionally, it should be noted that for the sake of description, only parts related to the present application are shown in the drawings rather than all of the content. Before discussing the exemplary embodiments in more detail, it should be mentioned that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although the flowcharts describe the operations (or steps) as sequential processes, many of the operations can be implemented in parallel, concurrently, or simultaneously. In addition, the order of the operations can be rearranged. When the operations are completed, the process can be terminated, but there may also be additional steps not included in the drawings. The process can correspond to a method, function, procedure, subroutine, subprogram, etc.

[0017] 1. Traditional technologies a) Manual measurement and analysis Traditional manual measurement and analysis methods for torque values are mainly based on two-dimensional images or physical models. Usually, the angle between the long axis of the tooth and the reference plane (such as the maxillary plane or the true horizontal plane) is determined by observing and measuring the side of the clinical crown of the tooth. The specific operations include using an intraoral scanner, a plaster model, or an X-ray image. After manually drawing the long axis of the tooth, the torque value is calculated through an angle measurement tool.

[0018] The characteristics of this method lie in relying on the experience and subjective judgment of the operator. The measurement accuracy may be limited by the model angle, image distortion, or measurement tools. In addition, due to the three-dimensional nature of the tooth long axis in different directions, two-dimensional images may lead to measurement results inconsistent with the actual three-dimensional positions. Although the manual measurement method is relatively simple and easy to implement, its accuracy and repeatability in the analysis of complex cases are insufficient and have gradually been replaced by more advanced methods such as CBCT combined with AI technology.

[0019] b) Three-dimensional model To overcome the constraints of two-dimensional projection measurement methods, three-dimensional model technology has gradually been introduced into the measurement of tooth gravity values. By reconstructing three-dimensional tooth models through CBCT (Cone Beam Computed Tomography) technology, doctors can analyze teeth more intuitively. However, most current applications of three-dimensional models still require doctors to manually calibrate key points of teeth, such as cusp tips and root centers, and finally calculate the long-axis angle of the teeth. Although this method improves the observability of measurement, the operation steps are cumbersome and repetitive, and it is still easily affected by human factors. Therefore, although three-dimensional model-assisted technology has improved the measurement accuracy to a certain extent, its efficiency and consistency issues need to be further optimized.

[0020] 2. Improved Methods Based on Digital Technology a) Digital Tooth Model Analysis With the popularization of digital technology, digital tooth models generated based on CBCT or optical scanning have been widely used in orthodontic treatment evaluation. With the help of professional software (such as OrthoAnalyzer, InVivo, and Dental), doctors can extract the long axis of teeth in digital models and calculate their inclination angles. These software usually provide a certain degree of automated analysis function, and by setting key parameters, basic measurements can be quickly completed. However, in complex cases, these methods have certain limitations. For example, in the face of abnormal tooth morphology or data loss, the accuracy of their analysis results is often difficult to guarantee. In addition, the operator needs to have a high level of professional technology. Especially when setting parameters and understanding the measurement logic, it is easy to affect the measurement results due to misoperation. These requirements limit the popularity and application scope of the technology to a certain extent.

[0021] b) Reference Plane Calculation When measuring the posture value, a standard reference plane needs to be selected, such as the maxillary plane (PP) or the normal occlusion plane. Adjusting the posture between the model and the reference plane is a key step in the measurement process, but this process often requires manual adjustment of the model's posture. The technical level of the operator may lead to inconsistencies in the planning results due to differences in experience and habits for operators with different adjustment precisions, thus affecting the reliability of the measurement. Therefore, how to achieve automated related planning and maintain the accuracy of the planning in complex scenarios is an important direction for the improvement of digital technology.

[0022] c) Tooth Long Axis Extraction The long axis of the tooth is a fundamental parameter for evaluating the movement value of the dental body, and its extraction process is directly related to the accuracy of the measurement results. Existing methods usually determine the long axis direction by identifying the geometric centerline of the tooth surface. This method performs well when dealing with teeth with regular shapes, but when dealing with complex or abnormal tooth shapes, the accuracy of the centerline may decrease significantly. Especially in the presence of data noise or tooth defects, the deviation of the extraction results will be further increased, affecting the reliability of subsequent measurements. Therefore, the long axis extraction technology needs to be more robust to adapt to complex and changing clinical scenarios.

[0023] 3. Preliminary Applications of Artificial Intelligence Technology a) Image Search Recently, the development of deep learning technology has brought new opportunities to oral medicine. Image segmentation technologies based on models such as U-Net have been widely used to process CBCT data, significantly improving the accuracy of tooth and bone segmentation. These technologies train on image data through topological neural networks and can quickly identify the boundaries and shapes of teeth. However, the capabilities of most existing models are limited to the overall differentiation of teeth from teeth, and it is difficult to achieve fine-grained segmentation of tooth shapes. Especially during extraction, further optimizing the segmentation model to enhance its fine-grained segmentation ability is an important direction for artificial intelligence technology in this field.

[0024] b) Three-dimensional Feature Display In addition to image segmentation, deep learning technology has also been used to process the three-dimensional point cloud data of teeth, automatically extracting geometric features through models such as PointNet or PointNet++. These models can analyze the spatial shape of teeth and quickly extract the long axis direction of teeth, providing a basis for the calculation of torque values. However, the generalization ability of these methods in complex scenarios still needs to be improved. Especially when dealing with data noise and abnormal points, their robustness and stability still need to be further verified.

[0025] c) Automated Torque Value Calculation Currently, some studies have attempted to combine CBCT images and artificial intelligence technology to develop an automated torque value measurement system. These systems use deep learning models to process the data and directly output the tilt angle and torque value of the tooth posture. However, since the research in this field is still in its infancy, there is still much room for improvement in the algorithm performance and clinical applications of existing systems. In addition, the standardization issue of different data sources also hinders the popularization and application of the system.

[0026] The disadvantages of the above solutions are respectively: 1. Manual operation steps are cumbersome and inefficient The existing technology still relies on a large amount of manual operations in the measurement of tooth torque values, such as the structure of the model, the calibration of key points of the gear, and the extraction of the tooth long axis. Since these steps require high professional skills of the operators, the whole process not only runs slowly but also easily leads to a low task completion rate. This cumbersome process is extremely inefficient when dealing with a large number of cases, which not only increases the workload but also may lead to too long operation time and affect the timeliness of the diagnosis result. Therefore, how to simplify the process and improve the operation efficiency has become an urgent problem to be solved.

[0027] 2. Difficulty in ensuring accuracy and consistency Whether it is two-dimensional measurement or three-dimensional model-assisted technology, the accuracy and consistency of its results are easily affected by various factors. Taking the layout of the reference plane as an example, the technical levels and experiences of different operators may lead to different measurement results. In addition, during the extraction of the tooth long axis, the existing algorithms malfunction due to the complexity and abnormalities of tooth morphology (such as tooth defects or dental crowding), and the accuracy of the extraction results often fails to meet the clinical requirements. These problems make it difficult to ensure the repeatability and reliability of the torque value measurement results, thus having a potential impact on the formulation of the orthodontic cycle plan.

[0028] 3. Insufficient automation Although some artificial intelligence technologies have achieved image segmentation and the realization of three-dimensional structures, the complete automated sequence value calculation system is completely not in line with the clinic. On the one hand, this is due to the insufficient generalization ability of the existing models in dealing with complex scenarios; on the other hand, it also reflects the imperfect integration between different technologies. For example, from the acquisition of CBCT data to the reconstruction, calculation, tooth long axis elevation, and torque value of the model, this complete process requires the improvement of an independent system, and the overall automation level is low. This not only increases the difficulty of use but also restricts the popularization and application of the technology.

[0029] The accurate measurement of tooth torque value is of great significance in orthodontic treatment. Although the existing technologies have made certain progress in terms of efficiency and accuracy, there are still problems such as cumbersome manual operations, difficulty in ensuring accuracy, insufficient automation, and lack of universality in applicability. The embodiment of the present application provides a method for measuring the labial (buccal)-lingual torque value of maxillary molars based on improved ResNET, aiming to solve the problems of insufficient accuracy, low efficiency, and low automation in the existing technologies. To achieve this goal, the embodiment of the present application starts from three aspects: improving measurement accuracy and efficiency and optimizing clinical treatment decision support. In terms of improving measurement accuracy and efficiency, through the measurement model of the posterior tooth torque value based on the crown-alveolar bone coupling relationship, the problems of insufficient accuracy and low efficiency in traditional technologies are solved, the accuracy and consistency of the measurement results are ensured, and at the same time, manual workpieces are reduced and work efficiency is improved; in terms of optimizing clinical treatment decision support, through the tooth long axis positioning system based on the improved ResNET model and the correction model of the tooth long axis positioning system with head position calibration, real-time and personalized treatment suggestions and treatment effects are provided for orthodontists, promoting the digital and standardized development of oral treatment.

[0030] The method for measuring the buccal-lingual torque value of maxillary molars based on improved ResNET provided in the embodiment can be executed by a device for measuring the buccal-lingual torque value of maxillary molars based on improved ResNET. The device for measuring the buccal-lingual torque value of maxillary molars based on improved ResNET can be implemented in a software and / or hardware manner and integrated in a device for measuring the buccal-lingual torque value of maxillary molars based on improved ResNET. Among them, the device for measuring the buccal-lingual torque value of maxillary molars based on improved ResNET can be a device such as a computer.

[0031] Figure 1 This is a flowchart of a method for measuring the buccal-lingual torque value of maxillary molars based on improved ResNET provided by the embodiment of the present application. Refer to Figure 1 , the method includes the following steps: 100. Obtain the three-dimensional images of teeth and alveolar bone, perform preprocessing, and perform three-dimensional reconstruction on the preprocessed images.

[0032] Specifically, use a cone beam computed tomography (CBCT) device to obtain the three-dimensional images of teeth and alveolar bone; for the three-dimensional images, apply median filtering or Gaussian filtering technology to remove the noise introduced during the scanning process, use the contrast-limited adaptive histogram equalization (CLAHE) technology to enhance the contrast between teeth and alveolar bone, and extract the region of interest (ROI); use open-source tools to perform three-dimensional reconstruction on the preprocessed images to generate a visual model of teeth and alveolar bone; among them, the three-dimensional reconstruction includes image segmentation, feature extraction, and three-dimensional modeling.

[0033] 200. Locate the first tooth long axis based on the coupling relationship between the dental crown and the tooth root.

[0034] Specifically, use region growing or threshold segmentation algorithms to separate the dental crown and the tooth root; starting from the seed points through the region growing algorithm, gradually add adjacent pixels similar to the seed points until the entire tooth region is covered. Through threshold segmentation, the tooth is distinguished from the background based on the range of pixel values. Usually, histogram analysis needs to be performed first to determine the optimal threshold; use U-Net or its variant networks to automatically segment the tooth region and extract point cloud data from the segmented dental crown and tooth root regions. Among them, the point cloud data includes the three-dimensional coordinate information of the tooth.

[0035] Apply principal component analysis (PCA) to calculate the principal axis directions of the dental crown and the tooth root. PCA projects the data into a new coordinate system through linear transformation so that the variance on the principal component of the first coordinate axis is the largest to determine the principal axis direction; use the least squares method to fit the midlines of the dental crown and the tooth root; among them, find the best fit line by minimizing the sum of the squares of the errors, and connect the midlines of the dental crown and the tooth root to form the first tooth long axis; display the fitting result through the visualization interface; if it is found that the long axis deviates severely, adjust the position of the connection point between the dental crown and the tooth root.

[0036] 300. Locate the second tooth long axis based on the coupling relationship between the tooth and the alveolar bone.

[0037] Specifically, perform three-dimensional curvature analysis on the alveolar bone region, and identify the main boundaries of the alveolar bone by calculating the curvature of the alveolar bone surface; select the intersection point of the extension lines so that the focus is located inside the alveolar bone; use the direction vectors of the extension lines to calculate their bisecting direction, and the formula is: ; Among them, vector L1 and vector L2 are the direction vectors of two extension lines passing through the contact points between the tooth and the alveolar bone; the direction vectors are bisected to determine a direction that takes into account both the tooth and the alveolar bone; use the bisector as the second tooth long axis and record its spatial coordinates. Vector L1 and vector L2 are generated by vector addition and normalization to obtain the bisecting direction of the two extension lines, which is used to correct the tooth long axis positioning; improve the robustness of the long axis extraction, reduce manual intervention, and adapt to complex anatomical structures.

[0038] 400. Calculate the included angle between the first tooth long axis and the second tooth long axis and generate a new long axis.

[0039] Specifically, start from the direction vectors of the two tooth long axes. Among them, the vectors are obtained from the long axis data obtained by fitting, which define the direction of the long axis in the three-dimensional space; use the dot product formula to calculate the cosine value of the included angle between the two vectors, and the dot product formula is: ; Wherein, a and b are two vectors, and are the moduli of the vectors, and θ is the included angle between the vectors; The included angle θ is obtained through the arccosine function. This included angle represents the spatial relationship between two long axes and is an important parameter for evaluating tooth alignment and alveolar bone structure; A new long axis is generated using the angle bisector. The direction vector of the angle bisector can be obtained through the average value of the normalized sum of the two original long axis direction vectors, that is: ; The new long axis is recorded and visualized to match the actual anatomical structure; during the visualization process, the position and direction of the new long axis are adjusted.

[0040] 500. Calculate the buccolingual torque value of the maxillary molar and verify it.

[0041] Specifically, determine the included angle between the new long axis and the alveolar bone plane, and calculate the included angle through the dot product of vectors; wherein, the direction vector of the new long axis is dotted with the normal vector of the alveolar bone plane, and the obtained result is used to calculate the cosine value of the included angle, and then the included angle is obtained; calculate the horizontal and vertical offsets of the crown top and root bottom along the new long axis; determine the precise positions of the crown and root, and measure the offsets relative to the new long axis; the calculation formula for the buccolingual torque value of the maxillary molar is expressed as: ; Wherein, θ is the included angle between the new long axis and the alveolar bone plane, and Δy and Δx are the horizontal and vertical offsets of the crown top and root bottom along the new long axis, respectively.

[0042] Compare the calculated torque value with the normal anatomical reference value or the clinical orthodontic target value; if the measurement error is large, check the quality of the CBCT image or the accuracy of the segmentation model. Among them, the image quality check includes the evaluation of parameters such as resolution, noise level, and contrast, and the accuracy check of the segmentation model is to retrain or adjust the model parameters.

[0043] As described above, the embodiments of the present application acquire three-dimensional images of teeth and alveolar bone and perform preprocessing, and perform three-dimensional reconstruction on the preprocessed images; based on the coupling relationship between the crown and the root, position the first tooth long axis; based on the coupling relationship between the tooth and the alveolar bone, position the second tooth long axis; calculate the included angle between the first tooth long axis and the second tooth long axis and generate a new long axis; calculate the buccolingual torque value of the maxillary molar and verify it; through an automated process, accurate measurement and calculation of the torque value are achieved, improving effectiveness and accuracy.

[0044] Exemplarily, through CBCT images, based on the crown-root coupling relationship, take Figure 4The true vertical line shown forms the first type of tooth long axis positioning. This method takes into account the relationship between the teeth themselves and believes that the torque value is mainly formed by the deflection of the posterior teeth after long-term force. Figure 4 The Lnew shown forms the second tooth long axis positioning. This method takes into account the relationship between the tooth and the alveolar bone, and believes that the torque value is mainly formed by the deflection formed by the reverse force of the alveolar bone on the tooth. At the same time, the angle between the two tooth long axes is calculated. On the premise of ensuring the same side, the bisector of the new angle is used as the new tooth long axis. This deviation balancing method takes into account the equal influence of the force on the tooth and the force on the alveolar bone.

[0045] The specific implementation process can be divided into the following parts: Acquisition and preprocessing of CBCT image data: In the process of dental 3D image analysis, the first step is the data acquisition stage. Using cone beam computed tomography (CBCT) equipment, detailed 3D images of the user's teeth and alveolar bones can be obtained. In order to ensure the clarity and low noise of the image, the setting of scanning parameters is crucial, including the resolution is usually set to 0.3 mm or higher to capture subtle structures; the voltage is adjusted between 70-120kVp to obtain appropriate X-ray penetration; the current is controlled at 5-10mA to balance image quality and user radiation dose. The precise setting of these parameters is crucial for subsequent diagnosis and treatment planning.

[0046] The second is the image preprocessing stage. First, in order to remove the noise introduced during the scanning process, median filtering or Gaussian filtering techniques are applied. These methods can effectively reduce noise while retaining edge information. Subsequently, in order to enhance the contrast between the teeth and the alveolar bone, adaptive histogram equalization (CLAHE) technology is used. This is a local contrast enhancement method that can significantly improve the clarity of the boundary area. Finally, the region of interest (ROI) is extracted. This step manually or automatically marks the tooth area and removes irrelevant background to reduce interference for subsequent analysis.

[0047] Finally, the 3D reconstruction phase begins. Using open source tools such as ITK-SNAP or 3D Slicer, CBCT images are reconstructed in 3D to generate visual models of teeth and alveolar bones. This step involves complex image processing and computer vision technologies, including image segmentation, feature extraction, and 3D modeling. The resulting 3D model can intuitively display the structure of teeth and alveolar bones, providing strong visual support for clinical diagnosis and treatment. By implementing these technologies in detail, accurate 3D information can be extracted from the original CBCT images, laying a solid foundation for subsequent dental analysis and treatment planning.

[0048] The first type of tooth long axis positioning: based on the crown-root coupling relationship, tooth segmentation: Traditional methods: Region growing or threshold segmentation algorithms are used to separate the crown and root of the tooth. The region growing algorithm starts from a seed point and gradually adds neighboring pixels similar to the seed point until the entire tooth region is covered. Threshold segmentation distinguishes the tooth from the background based on the range of pixel values, usually requiring prior histogram analysis to determine the optimal threshold.

[0049] Deep learning methods: U-Net or its variant networks are used, which are trained on a large amount of annotated dental image data to automatically segment the tooth region. The structure of U-Net includes a contracting path for capturing context information and a symmetric expanding path for precisely locating the tooth boundary, thus achieving high-precision automatic segmentation.

[0050] Centerline fitting: First, point cloud data is extracted from the segmented crown and root regions of the tooth, which contains the three-dimensional coordinate information of the tooth.

[0051] Next, principal component analysis (PCA) is applied to calculate the principal axis directions of the crown and root of the tooth. PCA projects the data into a new coordinate system through linear transformation, making the variance on the first coordinate axis (principal component) the largest, thereby determining the principal axis direction.

[0052] Finally, the least squares method is used to fit the midlines of the crown and root of the tooth. This method finds the best fit line by minimizing the sum of the squares of the errors and connects the midlines of the crown and root to form the first tooth long axis.

[0053] Inspection and correction: The fitting result is displayed through a visualization interface, and professionals can check whether the long axis deviates from the actual anatomical form to ensure the accuracy of the fitting.

[0054] If it is found that the long axis deviates severely, the system allows manual adjustment of the position of the connection point between the crown and root of the tooth. This interactive correction process ensures the accuracy of the final result, enabling the fitting of the tooth long axis to truly reflect the anatomical structure of the tooth.

[0055] Location of the second tooth long axis: Based on the coupling relationship between the tooth and the alveolar bone, extraction of the extended line of the alveolar bone: First, three-dimensional curvature analysis is performed on the alveolar bone region. This step identifies the main boundaries of the alveolar bone by calculating the curvature of the alveolar bone surface. Curvature analysis can reveal the geometric characteristics of the alveolar bone surface and help determine the edges of the alveolar bone. Then, using the contact points between the tooth and the alveolar bone, the extended line of the alveolar bone is fitted. This usually involves linear regression or the least squares method to determine the best straight line passing through the contact points. Finally, the intersection points of the extended lines are selected, ensuring that the focus is located inside the alveolar bone. This step requires precise geometric calculations to ensure that the intersection points are not only geometrically feasible but also anatomically meaningful.

[0056] Bisector calculation: Using the direction vector of the extension line, calculate its bisecting direction. The formula is: ; Among them, the direction vector is bisected to determine a direction that takes into account both the teeth and the alveolar bone. The bisector is used as the second tooth long axis, and its spatial coordinates are recorded. This step involves parameterizing the bisector in three-dimensional space, usually using parametric equations to describe the position and direction of the line in space.

[0057] Calculate the included angle between the two tooth long axes and generate a new long axis. Included angle calculation: First, it is necessary to start from the direction vectors of the two tooth long axes. These vectors can be obtained from the long axis data obtained by fitting, and they define the direction of the long axis in three-dimensional space. Then, use the dot product formula to calculate the cosine value of the included angle between the two vectors. The dot product formula is , where a and b are two vectors, and are the magnitudes of the vectors, and θ is the included angle between the vectors. Finally, the included angle θ is obtained through the inverse cosine function (arccos). This included angle represents the spatial relationship between the two long axes and is an important parameter for evaluating tooth alignment and alveolar bone structure.

[0058] New long axis generation: Use the bisector of the included angle to generate a new long axis. The direction vector of the bisector can be obtained by taking the average of the normalized sum of the two original long axis direction vectors, that is: ; Record and visualize the new long axis to ensure consistency with the actual anatomical structure. This step usually involves drawing the new long axis in three-dimensional visualization software and comparing it with the original image data to verify the accuracy and practicality of the new long axis. During the visualization process, the position and direction of the new long axis can be adjusted to ensure that it coincides with the actual anatomical structure of the teeth and alveolar bone.

[0059] Measurement and verification of torque value. Torque value calculation: First, determine the included angle between the new long axis and the alveolar bone plane. This included angle can be calculated through the vector dot product, where the direction vector of the new long axis is dotted with the normal vector of the alveolar bone plane, and the result is used to calculate the cosine value of the included angle, and then the included angle is obtained. Then, calculate the horizontal and vertical offsets (Δy and Δx) of the top of the tooth crown and the bottom of the tooth root along the new long axis. This usually involves positioning the tooth model in three-dimensional space, determining the precise positions of the tooth crown and the tooth root, and measuring their offsets relative to the new long axis. The calculation formula for the torque value can be expressed as: ; where θ is the angle between the new major axis and the alveolar bone plane, and Δy and Δx are the horizontal and vertical offsets of the crown top and root bottom along the new major axis, respectively.

[0060] Result verification: Compare the calculated torque value with the normal anatomical reference value or the clinical correction target value. This step is to verify whether the measured value is within the acceptable range and whether it meets the clinical correction target. If the measurement error is large, it is necessary to check the CBCT image quality or the accuracy of the segmentation model. The image quality check may include the evaluation of parameters such as resolution, noise level, and contrast. The accuracy check of the segmentation model may involve retraining or adjusting the model parameters to ensure the accurate segmentation of the tooth area.

[0061] In addition, a tooth major axis positioning system based on the improved ResNET model: During the training process of CBCT (Cone Beam Computed Tomography) images, in order to meet the real-time computing requirements of on-site deployment, the improved ResNet18 architecture (ResNet18-s) is adopted. By introducing the SE (Squeeze-and-Excitation) mechanism, the computing time is optimized. The SE module adaptively assigns weights to each feature channel, enhancing the network's sensitivity to key features while reducing the dependence on irrelevant information. It is embedded in the feature extraction stage of the network to improve the model's expressive ability.

[0062] Set the input feature map. The SE module first performs global average pooling (Global Average Pooling, GAP) on the spatial information of each channel to calculate the global statistics of each channel: ; Then use two fully connected layers and a non-linear activation function to model the dependence relationship between channels and generate the weight ratio s of each channel: ; where, is the weight matrix of the first fully connected layer, used to compress the channel dimension and reduce the computing cost. is the weight matrix of the second fully connected layer, used to restore the channel dimension. is the Sigmoid activation function, used to normalize the channel weights.

[0063] Finally, use the channel weight ratio s to re-weight the original feature map to obtain the output feature map: ; The residual block of ResNet consists of a main branch and a bypass connection. The main branch contains two convolutional layers and their corresponding batch normalization (BN) and activation function (ReLU). The bypass connection directly adds the input feature map to the output of the main branch: ; In ResNet18-s, the SE module directly replaces the convolutional layer and its activation process in the main branch while retaining the bypass connection. The replaced expression is: ; where is the feature weighted by the SE module.

[0064] Since the calculation of the SE module only involves the weight distribution between channels, the spatial complexity is much lower than the original convolution operation. Therefore, replacing the specified layer with the SE module in ResNet18 reduces the computational amount while enhancing the model's ability to focus on key features. Specifically for ResNet18-s: Layers 3, 4, and 5 are combined into one SE module; Layers 8, 9, 10, and 11 are combined into one SE module; Layers 12, 13, 14, and 15 are combined into one SE module. By reducing redundant convolution operations, the computational overhead is significantly reduced.

[0065] After training is completed, the following formula is used to calculate the torque value of each person during inference: ; where: are the torque values output by the anterior tooth model, canine tooth model, and posterior tooth model respectively; is the torque value of the global model.

[0066] The tooth long axis positioning system correction model combined with head position calibration: The key to tooth long axis positioning is the head position level. Based on CBCT images, the median palatine suture, anterior nasal spine point - posterior nasal spine point (ANS - PNS), and nasal septum are determined as three calibration baselines. As Figure 5 shown, the above tooth long axis positioning is pre - corrected. The correction model uses the Unet framework, marks the left - right segmentation parts of the three calibration baselines, and determines the imbalance ratio according to the area ratio. The three ratio coefficients are input into an MLP to calculate the 3D rotation angle. According to the calculated value, the rotation angle of the CBCT is adjusted until the imbalance ratio is 1. The head position calibration is completed.

[0067] The specific implementation process can be divided into the following parts: data collection and preliminary pre - processing, CBCT image acquisition: When collecting three-dimensional imaging data of the user's head, using a cone beam computed tomography (CBCT) device is a crucial step, which can capture detailed structures including the skull and teeth. To ensure the clear visibility of fine structures such as the nasal septum, median palatal suture, and ANS-PNS baseline in the image, the scanning resolution needs to be set at 0.3 mm or higher. Such a high-resolution setting is crucial for subsequent diagnosis and treatment planning because it provides sufficient details for professional analysis.

[0068] Image preprocessing: When processing CBCT data, first, the Gaussian filtering technique is used to denoise the image to eliminate artifacts and noise that may affect the image quality. This step is crucial for improving the accuracy of subsequent analysis. Then, through the image normalization step, the gray value range of all imaging data is adjusted to the standard interval of [0, 1]. This can enhance the consistency of data among different users and facilitate comparative analysis. Finally, the region of interest (ROI), such as key structures like the nasal cavity, median palatal suture, and ANS-PNS, is marked manually or automatically. This cropping step not only reduces the data volume during data processing but also significantly improves the efficiency of subsequent processing and analysis. These three steps together constitute the core process of image preprocessing, providing a high-quality imaging data foundation for subsequent diagnosis and treatment planning.

[0069] Extraction of the calibration baseline, extraction of the median palatal suture: In the preliminary positioning stage, first, a deep learning segmentation model, such as the U-Net architecture, is used to train the labeled data to accurately extract the median palatal suture region from the CBCT image. The U-Net model is widely used in medical image processing for its excellent image segmentation ability and can identify and segment complex biological structures. After segmentation, further processing is carried out through a centerline extraction algorithm on the segmentation result to obtain the central axis of the median palatal suture. This step is crucial because it provides an accurate anatomical reference line for subsequent analysis. Subsequently, in the optimization and smoothing section, polynomial fitting or spline interpolation techniques are used to smooth the central axis. These mathematical tools can effectively interpolate the points on the central axis, reducing fluctuations caused by segmentation errors or noise, thus ensuring the continuity and smoothness of the central axis and providing a more accurate and reliable data basis for subsequent measurement and analysis. These technical implementation details together ensure the accuracy and usability of the central axis of the median palatal suture extracted from CBCT data, providing strong technical support for clinical applications.

[0070] Extraction of the ANS-PNS line: In oral and maxillofacial image analysis, the accurate identification of the anterior nasal spine point (ANS) and the posterior nasal spine point (PNS) is crucial. First, the palatal bone region is extracted through a deep learning segmentation model, and a shape-based algorithm, such as curvature analysis, is used to detect the ANS at the front end of the palatal bone. Then, the PNS is determined at the posterior part of the palatal bone through an edge detection algorithm, which may involve extreme point search and gradient analysis to ensure its accuracy. Finally, by calculating the straight-line vector between the ANS and the PNS and using the linear interpolation method, these two points are connected to form a calibration baseline, providing an accurate reference for maxillofacial structure analysis. The combination of these techniques ensures the accuracy of the key anatomical points extracted from the image data and the reliability of the calibration baseline, providing strong technical support for clinical diagnosis and treatment planning.

[0071] Septum extraction: In the automatic segmentation stage, first, the U-Net deep learning model is applied to accurately segment the nasal cavity region in the CBCT image. This model can identify and extract the central plane of the nasal septum. U-Net is widely used due to its excellent performance in image segmentation tasks. Its structure includes a contracting path for capturing context information and a symmetric expanding path for precise localization. After obtaining the central plane of the nasal septum, the centerline of the nasal septum is then extracted through 3D skeletonization, which is a morphological thinning process. This step can highlight the central axis in the three-dimensional structure, providing an accurate geometric reference for subsequent analysis. In the symmetry analysis stage, the system automatically compares the volumes on both sides of the nasal cavity to verify the accuracy of the nasal septum segmentation. This usually involves calculating the difference in volumes on both sides and comparing it with a preset threshold. If a significant deviation is found, the system will automatically correct it or remind the operator to conduct a manual inspection through the user interface to ensure the accuracy of the analysis results. This process may involve machine learning algorithms for identifying and correcting biases in segmentation or using geometric algorithms to adjust the position of the centerline to ensure that the centerline of the nasal septum is geometrically centered, thus providing reliable analysis results for clinical applications.

[0072] Baseline segmentation and unbalanced ratio calculation: In the baseline segmentation process, first, using the output results of the deep learning segmentation model, the three baselines of the palatal raphe, ANS-PNS, and nasal septum are split into two along their centerlines, and the regions on the left and right sides are respectively extracted. Then, by calculating the number of pixels or volume of each side region, the segmentation areas AL and AR on the left and right sides are obtained. Specifically, for each baseline, the formula for calculating the unbalanced ratio is: ; Among them, ∣AL−AR∣ represents the absolute difference in area between the left and right sides, and AL+AR is the total area of the two sides. The value range of this proportionality coefficient R is between 0 and 1. The closer the value is to 0, the more balanced the left and right sides are, and the closer the value is to 1, the higher the degree of imbalance.

[0073] Next, calculate the imbalance ratio for the three baselines of the palatal suture, ANS-PNS, and nasal septum respectively, obtaining three proportionality coefficients R1, R2, and R3. These coefficients can provide clinicians with a quantitative reference for the degree of nasal septum deviation, helping them evaluate the necessity and plan for surgical correction. For example, if any of the values of R1, R2, or R3 exceeds a preset threshold, it may indicate a significant deviation of the nasal septum and further medical intervention is required.

[0074] Implementation of the head position calibration model, model selection: As Figure 6 shown, use a calibration correction model based on a multi-layer perceptron (MLP) to input the imbalance ratios of the three baselines and output the 3D rotation angles required for calibration.

[0075] MLP structure design: When designing the multi-layer perceptron (MLP) structure to predict the 3D rotation angles, first construct an input layer that receives the three proportionality coefficients R1, R2, and R3 obtained from the baseline segmentation and imbalance ratio calculation steps. These coefficients serve as the input features of the network and are used to predict the rotation angles required for nasal septum correction. For 3D rotation, a rotation matrix can be used to represent the rotation around the X, Y, and Z axes. For example, the rotation matrix for rotating an angle θ around the Z axis is: ; Next, the MLP structure contains two hidden layers, each using a fully connected network. The fully connected layer can map the input features to a higher-dimensional space, enabling the network to learn the complex relationships between the input data. In these two hidden layers, the ReLU (Rectified Linear Unit) is used as the activation function, which can introduce non-linearity, enhance the expression ability of the network, and avoid the gradient vanishing problem, making the network training more efficient.

[0076] Finally, the output layer of the MLP is designed to output three values, corresponding to the rotation angles α, β, and γ around the X, Y, and Z axes respectively. These rotation angles are predicted based on the input imbalance ratio coefficients and are used to guide the actual nasal septum correction surgery. By minimizing the error between the predicted rotation angles and the actual rotation angles required for correction, the MLP model can be trained to accurately predict the 3D rotation angles required for correcting the nasal septum deviation.

[0077] The design of the entire MLP structure aims to automatically provide precise guidance for septoplasty by learning the relationship between the unbalanced ratio of the baseline segmentation and the corrective rotation angles. This machine learning-based method can improve the accuracy and efficiency of the surgery and reduce the dependence on manual experience.

[0078] Model training: In the model training stage, the preparation of the dataset is carried out first. This includes collecting a series of labeled CBCT image data and calculating the actual required rotation angles (α, β, γ) for each sample, which will be used as the labels for model training. These label data are usually manually labeled by experts or obtained through precise medical measurements, ensuring the accuracy and reliability of the training data.

[0079] Next, the loss function of the model is defined as the mean squared error (MSE), which measures the average of the squares of the differences between the model's predicted values and the true values. Using MSE as the loss function helps to optimize the model parameters, making the predicted rotation angles more accurately match the actual rotation angles.

[0080] In the selection of the optimizer, the Adam optimization algorithm is adopted, which is an adaptive learning rate optimization algorithm and is widely used due to its high efficiency in dealing with non-convex optimization problems. The initial learning rate is set to 1e-4, which is a commonly used starting value and can be adjusted according to the actual situation during the training process.

[0081] During the training process, the dataset is divided into multiple batches. Forward propagation is performed on each batch of data to calculate the error between the predicted rotation angles and the true angles, and then the error is backpropagated through the backpropagation algorithm to update the weights and biases of the model. This process is repeated continuously until the performance of the model on the validation set no longer improves or reaches the preset number of iterations.

[0082] Through such a training process, the model can learn how to accurately predict the required 3D rotation angles based on the input ratio coefficients R1, R2, R3 to calibrate the deviation of the nasal septum. This deep learning-based method can automatically extract features from complex image data and predict accurate rotation angles, providing strong assistance for clinical surgery.

[0083] 5) Head position calibration: During the head position calibration process, first, the pre-trained multi-layer perceptron (MLP) model is used to infer the input imbalance ratio coefficients R1, R2, and R3. Specifically, these ratio coefficients are used as the input of the model, and through forward propagation, the output of the model, that is, the predicted 3D rotation angles (α, β, γ), is calculated. These angles respectively correspond to the rotation amounts around the X, Y, and Z axes. This step involves applying the weights and biases of the model to the input data and calculating the final output result through activation functions and connections between layers.

[0084] After obtaining the predicted rotation angles, the pose of the CBCT image is adjusted according to these angles. Technically, this usually involves using 3D graphics processing libraries such as VTK or ITK to apply rotation transformations to the image data. Specifically, a rotation matrix is constructed, which is defined according to the predicted rotation angles α, β, γ, and then this rotation matrix is applied to each voxel or pixel of the CBCT image to achieve the spatial rotation of the image.

[0085] After the rotation adjustment, the imbalance ratios of the three baselines are recalculated to evaluate the calibration effect. If the imbalance ratio has not reached the target close to 1, that is, the volume difference between the left and right sides is still large, the above steps are repeated, and the MLP model is used for inference again, and the image pose is adjusted according to the new rotation angles. This process will be iterated until the imbalance ratios of the three baselines are all close to 1, indicating that the head position calibration is completed. At this time, the CBCT image achieves the required symmetry on the left and right sides.

[0086] The key to this process lies in the accuracy of the MLP model and the fine control of the iterative calibration, ensuring the gradual adjustment from the initial pose to the final pose, and finally achieving accurate head position calibration. By automating this process, the efficiency and accuracy of head position calibration can be significantly improved, providing high-quality image data for subsequent clinical analysis and treatment planning.

[0087] 6) Optimization of the long axis positioning of the posterior teeth after calibration In the calibrated CBCT image, the re-extraction of the long axis of the teeth is first carried out. Specifically, the first step is to use image processing and pattern recognition techniques such as the Hough transform or deep learning segmentation models to identify and extract the coupled part of the tooth crown and root, so as to obtain the first long axis of the tooth. The key to this step lies in accurately identifying the boundaries and directions of the teeth in order to accurately determine the position of the long axis. Immediately afterwards, by analyzing the coupled area between the teeth and the alveolar bone, similar image analysis techniques are used to recalculate the second long axis of the tooth, which requires an in-depth understanding of the anatomical structure of the teeth and the alveolar bone.

[0088] In the stage of calculating the corrected major axis angle, based on the calibrated tooth major axis data, the angle between the two major axes is recalculated. This usually involves vector operations such as dot product or cross product to determine the spatial relationship between the two axes. After calculating the angle, the angle bisector is further determined, which can be achieved through vector decomposition and rotation operations, so as to obtain a more accurate new major axis that represents the average direction of the tooth.

[0089] In the error verification stage, the angles of the tooth major axis before and after calibration are compared to verify the impact of head position calibration on the positioning of the tooth major axis. This step evaluates the calibration effect by comparing the angle changes before and after calibration. If the angle change is within the expected range, the calibration is considered successful. If the error exceeds the expected range, the output of the calibration model will be checked, including the accuracy of the predicted rotation angle and the correctness of the scale factors R1, R2, and R3 input into the model. This may involve retraining or adjusting the MLP model, or rechecking the input data to ensure that all data is accurate and error-free.

[0090] Through this series of technical implementation steps, the head position can be accurately calibrated, and the tooth major axis can be accurately positioned and measured.

[0091] As mentioned above, existing traditional motion control methods, such as two-dimensional cephalometry and three-dimensional model analysis, often rely on manual operations and empirical judgments. The measurement accuracy is increasingly affected by the operator's skills and additional factors. However, in the embodiments of this application, the three-dimensional data of teeth are automatically processed through deep learning technology, and the long axis of the teeth is accurately obtained by using AI algorithms and the displacement value is calculated, greatly reducing the manual work and providing higher accuracy and consistency. Through the automated process, the gravity value measurements of all users will be highly consistent, avoiding deviations caused by operation biases.

[0092] In traditional technologies, the posture actions and calculation processes usually require orthodontists to manually calibrate key points, adjust the reference layout, extract the long axis, etc., with a cumbersome workflow. However, in the embodiments of this application, the full-process automated operation, from the input of CBCT data to the calculation of the motion values, is automatically completed by the AI system throughout the process, greatly improving the work efficiency. Data upload and treatment plan confirmation can be carried out, reducing cumbersome manual operations, saving a large amount of time, and being able to process more data.

[0093] Tooth posture measurement methods usually rely on professional experience, but there may be problems in complex cases. For example, in cases where the tooth morphology is complex or there is dental crowding, it is difficult for traditional methods to ensure accuracy. However, in the embodiments of this application, through deep learning and artificial intelligence technologies, automated tooth segmentation and long axis extraction are realized, eliminating the huge impact and potential errors in manual operations, reducing the operation risk, and ensuring the accuracy and reliability of the measurement results.

[0094] Existing technologies often require manual processing when dealing with complex tooth shapes, malocclusions, and other special cases, which increases the workload and complexity. However, the system of the embodiments of the present application can automatically identify and process complex tooth shapes through a deep learning model, reducing doctor intervention, efficiently handling various complex cases, and improving the efficiency and versatility of the system.

[0095] Existing technologies often have problems such as data incompatibility or poor information transfer between multiple software tools, resulting in a cumbersome operation process and inability to analyze. However, the system of the embodiments of the present application seamlessly integrates modules such as data input, three-dimensional model reconstruction, long axis extraction, and motion calculation, providing a unified platform, reducing the complexity of data processing, and greatly improving the convenience of operation. The user interface is friendly, and doctors can easily get started and quickly complete the entire measurement process.

[0096] Based on the above embodiments, please refer to Figure 2 A measuring device for the buccolingual torque value of maxillary molars based on an improved ResNET provided by the embodiments of the present application, the measuring device for the buccolingual torque value of maxillary molars based on an improved ResNET specifically includes: an image acquisition module 201, a first positioning module 202, a second positioning module 203, an included angle calculation module 204, and a torque calculation module 205.

[0097] Among them, the image acquisition module 201 is used to acquire three-dimensional images of teeth and alveolar bone and perform preprocessing, and perform three-dimensional reconstruction on the preprocessed images; the first positioning module 202 is used to position the first tooth long axis based on the coupling relationship between the tooth crown and the tooth root; the second positioning module 203 is used to position the second tooth long axis based on the coupling relationship between the tooth and the alveolar bone; the included angle calculation module 204 is used to calculate the included angle between the first tooth long axis and the second tooth long axis and generate a new long axis; the torque calculation module 205 is used to calculate the buccolingual torque value of the maxillary molar and perform verification.

[0098] As described above, the embodiments of the present application acquire three-dimensional images of teeth and alveolar bone and perform preprocessing, and perform three-dimensional reconstruction on the preprocessed images; position the first tooth long axis based on the coupling relationship between the tooth crown and the tooth root; position the second tooth long axis based on the coupling relationship between the tooth and the alveolar bone; calculate the included angle between the first tooth long axis and the second tooth long axis and generate a new long axis; calculate the buccolingual torque value of the maxillary molar and perform verification; through an automated process, accurate measurement and calculation of the torque value are achieved, improving effectiveness and accuracy.

[0099] The measurement device for the buccolingual torque value of maxillary molars based on the improved ResNET provided by the embodiments of the present application can be used to execute the measurement method for the buccolingual torque value of maxillary molars based on the improved ResNET provided by the above embodiments, and has corresponding functions and beneficial effects.

[0100] The embodiments of the present application also provide a computer device, which can integrate the measurement device for the buccolingual torque value of maxillary molars based on the improved ResNET provided by the embodiments of the present application. Figure 3 It is a schematic structural diagram of a computer device provided by the embodiments of the present application. Refer to Figure 3 , this computer device includes: an input device 33, an output device 34, a memory 32, and one or more processors 31; the memory 32 is used to store one or more programs; when the one or more programs are executed by the one or more processors 31, the one or more processors 31 implement the measurement method for the buccolingual torque value of maxillary molars based on the improved ResNET provided by the above embodiments. Among them, the input device 33, the output device 34, the memory 32, and the processor 31 can be connected through a bus or other means, Figure 3 Taking connection through a bus as an example.

[0101] The processor 31 executes various functional applications and data processing of the device by running software programs, instructions, and modules stored in the memory 32, that is, implements the above-mentioned measurement method for the buccolingual torque value of maxillary molars based on the improved ResNET.

[0102] The above-provided computer device can be used to execute the measurement method for the buccolingual torque value of maxillary molars based on the improved ResNET provided by the above embodiments, and has corresponding functions and beneficial effects.

[0103] The embodiments of the present application also provide a storage medium containing computer-executable instructions, and the computer-executable instructions are used to execute a measurement method for the buccolingual torque value of maxillary molars based on the improved ResNET when executed by a computer processor. The measurement method for the buccolingual torque value of maxillary molars based on the improved ResNET includes: adding scheduling information through the scheduling system background, where the scheduling information includes point positions, paths, and action information; automatically generating a running path based on the starting point and target point of the robot and dispatching it to the robot; according to the real-time task status and real-time working status of the elevator, combining the real-time action information of the robot, adjusting the running path in real time, and controlling the robot to run.

[0104] Storage medium - Any of various types of memory devices or storage devices. The term "storage medium" is intended to include: installation media such as CD-ROMs, floppy disks or magnetic tape devices; computer device memory or random access memory such as DRAM, DDRRAM, SRAM, EDORAM, Rambus RAM, etc.; non-volatile memory such as flash memory, magnetic media (such as hard disks or optical storage); registers or other similar types of memory elements, etc. The storage medium may also include other types of memory or combinations thereof. Additionally, the storage medium may be located in a first computer device in which the program is executed, or may be located in a different second computer device that is connected to the first computer device via a network (such as the Internet). The second computer device may provide program instructions to the first computer for execution. The term "storage medium" may include two or more storage media that may reside in different locations (such as in different computer devices connected via a network). The storage medium may store program instructions (such as embodied as a computer program) executable by one or more processors.

[0105] Of course, for a storage medium containing computer-executable instructions provided in an embodiment of the present application, the computer-executable instructions are not limited to the method for measuring the buccolingual torque value of maxillary molars based on the improved ResNET as described above, and may also perform related operations in the method for measuring the buccolingual torque value of maxillary molars based on the improved ResNET provided in any embodiment of the present application.

[0106] The measuring device, storage medium, and computer device for the buccolingual torque value of maxillary molars based on the improved ResNET provided in the above embodiments may execute the method for measuring the buccolingual torque value of maxillary molars based on the improved ResNET provided in any embodiment of the present application. For technical details not described in detail in the above embodiments, reference may be made to the method for measuring the buccolingual torque value of maxillary molars based on the improved ResNET provided in any embodiment of the present application.

[0107] The embodiments of the present application also provide a computer program product. The methods described in the various embodiments of the present application may be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented using software, it may be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer programs or instructions. When the computer program or instructions are loaded and executed on a computer, the processes or functions described in the various embodiments of the present application are executed in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, a network device, a user device, a core network device, an OAM (Open Application Model), or other programmable devices.

[0108] The above are only the preferred embodiments of the present application and the technical principles applied. The present application is not limited to the specific embodiments described herein. Various obvious changes, re-adjustments, and substitutions that can be made by those skilled in the art will not depart from the protection scope of the present application. Therefore, although the present application has been described in more detail through the above embodiments, the present application is not limited to the above embodiments. Without departing from the concept of the present application, it may also include more other equivalent embodiments, and the scope of the present application is determined by the scope of the claims.

Claims

1. A method for measuring the buccolingual torque value of maxillary molars based on improved ResNET, characterized in that, The method includes the following steps: Obtain three-dimensional images of teeth and alveolar bone and perform preprocessing, and perform three-dimensional reconstruction on the preprocessed images; Based on the coupling relationship between the crown and the root, locate the first tooth long axis; Based on the coupling relationship between the tooth and the alveolar bone, locate the second tooth long axis; Calculate the angle between the first tooth long axis and the second tooth long axis and generate a new long axis; Calculate the buccolingual torque value of the maxillary molar and verify it.

2. The method for measuring the buccolingual torque value of the maxillary molar based on the improved ResNET according to claim 1, wherein The obtaining of three-dimensional images of teeth and alveolar bone and performing preprocessing, and performing three-dimensional reconstruction on the preprocessed images includes: Use a cone beam computed tomography (CBCT) device to obtain three-dimensional images of teeth and alveolar bone; For the three-dimensional images, apply median filtering or Gaussian filtering technology to remove the noise introduced during scanning, use the contrast-limited adaptive histogram equalization (CLAHE) technology to enhance the contrast between teeth and alveolar bone, and extract the region of interest (ROI); Use open source tools to perform three-dimensional reconstruction on the preprocessed images to generate a visualization model of teeth and alveolar bone; among them, three-dimensional reconstruction includes image segmentation, feature extraction and three-dimensional modeling.

3. The method for measuring the buccolingual torque value of the maxillary molar based on the improved ResNET according to claim 1, wherein The locating of the first tooth long axis based on the coupling relationship between the crown and the root includes: Adopt region growing or threshold segmentation algorithm to separate the crown and the root; starting from the seed point through the region growing algorithm, gradually add adjacent pixels similar to the seed point until the entire tooth area is covered, and through threshold segmentation, the tooth is distinguished from the background based on the range of pixel values, usually requiring prior histogram analysis to determine the optimal threshold; Use U-Net or its variant network to automatically segment the tooth area; Extract point cloud data from the segmented crown and root regions, where the point cloud data includes the three-dimensional coordinate information of the teeth; Apply principal component analysis (PCA) to calculate the principal axis directions of the crown and the root. PCA projects the data into a new coordinate system through linear transformation so that the variance on the principal component of the first coordinate axis is the largest to determine the principal axis direction; Use the least squares method to fit the midlines of the crown and the root; among them, the best fit line is found by minimizing the sum of the squares of the errors, and the midlines of the crown and the root are connected to form the first tooth long axis; Display the fitting result through a visualization interface; If it is found that the long axis deviates severely, adjust the position of the connection point between the crown and the root.

4. The method for measuring the buccolingual torque value of the maxillary molar based on the improved ResNET according to claim 1, wherein, The locating of the second tooth long axis based on the coupling relationship between the tooth and the alveolar bone includes: Perform three-dimensional curvature analysis on the alveolar bone region to identify the main boundaries of the alveolar bone by calculating the curvature of the alveolar bone surface; Select the intersection point of the extension lines so that the focus is located inside the alveolar bone; Use the direction vectors of the extension lines to calculate its bisecting direction, and the formula is: ； Among them, vector L1 and vector L2 are the direction vectors of two extension lines passing through the contact points between the tooth and the alveolar bone; the direction vectors are bisected to determine a direction that takes into account both the tooth and the alveolar bone; the bisector is used as the second tooth long axis, and its spatial coordinates are recorded.

5. The method for measuring the buccolingual torque value of the maxillary molar based on the improved ResNET according to claim 1, wherein The calculating of the angle between the first tooth long axis and the second tooth long axis and generating a new long axis includes: Starting from the direction vectors of the two tooth long axes, where the vectors are obtained from the long axis data obtained by fitting and define the direction of the long axis in three-dimensional space; Use the dot product formula to calculate the cosine value of the angle between the two vectors. The dot product formula is: ； where a and b are two vectors, and are the magnitudes of the vectors, and θ is the angle between the vectors; Obtain the angle θ through the inverse cosine function. This angle represents the spatial relationship between the two long axes and is an important parameter for evaluating tooth alignment and alveolar bone structure; Generate a new long axis using the angle bisector. The direction vector of the angle bisector can be obtained by taking the average of the normalized sum of the two original long axis direction vectors, that is: ； Record and visualize the new long axis to match the actual anatomical structure; among them, during the visualization process, adjust the position and direction of the new long axis.

6. The method for measuring the buccolingual torque value of the maxillary molar based on the improved ResNET according to claim 1, wherein The calculation of the buccolingual torque value of the maxillary molar and its verification include: Determine the angle between the new long axis and the alveolar bone plane, and calculate the angle through vector dot product; among them, the direction vector of the new long axis is dotted with the normal vector of the alveolar bone plane, and the obtained result is used to calculate the cosine value of the angle, and then the angle is obtained; Calculate the horizontal and vertical offsets of the crown top and root bottom along the new long axis; Determine the exact positions of the crown and root, and measure the offsets relative to the new long axis; The calculation formula for the buccolingual torque value of the maxillary molar is expressed as: ； where θ is the angle between the new long axis and the alveolar bone plane, and Δy and Δx are the horizontal and vertical offsets of the crown top and root bottom along the new long axis respectively; Compare the calculated torque value with the normal anatomical reference value or the clinical orthodontic target value; If the measurement error is large, check the quality of the CBCT image or the accuracy of the segmentation model. Among them, the image quality check includes the evaluation of parameters such as resolution, noise level, and contrast, and the accuracy check of the segmentation model is to retrain or adjust the model parameters.

7. A measuring device for the buccolingual torque value of maxillary molars based on an improved ResNET, characterized in that, Including: An image acquisition module for acquiring three-dimensional images of teeth and alveolar bone, preprocessing them, and performing three-dimensional reconstruction on the preprocessed images; A first positioning module for positioning the first tooth long axis based on the coupling relationship between the crown and the root; A second positioning module for positioning the second tooth long axis based on the coupling relationship between the tooth and the alveolar bone; An angle calculation module for calculating the angle between the first tooth long axis and the second tooth long axis and generating a new long axis; A torque calculation module for calculating the buccolingual torque value of the maxillary molar and performing verification.

8. A computer device, characterized in that, Including: A memory and one or more processors; The memory is used to store one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement a method for measuring the buccolingual torque value of the maxillary molar based on the improved ResNET as described in any one of claims 1-6.

9. A storage medium containing computer-executable instructions, characterized in that, The computer-executable instructions are used to execute a method for measuring the buccolingual torque value of the maxillary molar based on the improved ResNET as described in any one of claims 1-6 when executed by a computer processor.

10. A computer program product, characterized in that, The computer program product includes instructions that, when executed by a computer, cause the computer to implement the method according to any one of claims 1-6.

Images