Equipment for analyzing morphologies of mandible bodies with different vertical bone surface types based on cone beam CT (Computed Tomography) data and application

Abstract

The invention discloses equipment for analyzing different vertical bone surface type lower jawbone body morphologies based on cone beam CT data and application, and relates to the technical field of stomatology. The image preprocessing module is connected with the image acquisition module and used for processing the original data to reduce image artifacts, and the image preprocessing module adopts a metal artifact correction algorithm based on projection data preprocessing and a motion artifact correction algorithm based on image registration; the image reconstruction module is connected with the image preprocessing module and used for reconstructing the processed image data into a three-dimensional image; a data analysis module; a display and interaction module; in the CBCT-based analysis technology image preprocessing stage, a motion artifact correction algorithm accurately distinguishes physiological motion from abnormal motion by means of a high-precision real-time motion monitoring device, preprocessing efficiency and quality are improved, and image errors are reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of oral medicine, and in particular to a mandibular body morphology analysis device based on cone beam CT data in different vertical bone face types and its application. Background Art

[0002] The morphological analysis of mandibular bodies in different vertical bone facial types using cone beam CT data is to use the three-dimensional image data of mandibles obtained by cone beam CT to conduct detailed research and comparison of various morphological characteristics of mandibular bodies in different vertical bone facial types (high angle type, average angle type, low angle type). This analysis helps orthodontists and oral and maxillofacial surgeons to better understand the morphological characteristics of mandibles in different bone facial types, and provide a basis for clinical diagnosis and treatment plan formulation (such as whether to extract teeth in orthodontic treatment, the direction of application of correction force, and the design of oral and maxillofacial surgery plans, etc.).

[0003] In the field of stomatology, mandibular morphology analysis is critical for orthodontics, maxillofacial surgery, and the diagnosis and treatment of temporomandibular joint diseases. Traditional two-dimensional X-ray imaging is difficult to present its three-dimensional structure, which limits the accuracy of diagnosis and the rationality of treatment plans. Although cone beam CT can obtain three-dimensional images, the existing CBCT-based analysis technology still has many defects. In the image preprocessing stage, the distinction and compensation of motion artifacts are insufficient, affecting image quality and efficiency. During image reconstruction, the super-resolution algorithm does not properly handle noise.

[0004] In view of this, this application is hereby filed. Summary of the invention

[0005] The purpose of the present invention is to provide a device and application for analyzing the mandibular body morphology in different vertical directions of bone facial types based on cone beam CT data, so as to solve the problem mentioned in the above background technology that the existing CBCT-based analysis technology is insufficient in distinguishing and compensating motion artifacts in the image preprocessing stage, which affects the image quality and efficiency.

[0006] In order to solve the above technical problems, the present invention provides a mandibular body morphology analysis device and application based on cone beam CT data in different vertical bone facial types, including:

[0007] Image acquisition module: used to obtain cone beam CT raw data including the mandibular body. The image acquisition module adopts variable field of view scanning technology to adjust the scanning field of view according to analysis requirements, and is equipped with a precise dose control algorithm, which can automatically adjust the radiation dose according to the patient's body shape and scanning site; at the same time, the module introduces an artificial intelligence-assisted scanning parameter optimization unit. Before scanning, the patient is first subjected to a simple mandibular body appearance inspection and medical history inquiry, and this information is input into the unit. The unit combines a large amount of previous case data to automatically generate more accurate initial scanning parameters, including field of view range, tube current, tube voltage, etc., and establishes specific radiation dose control models for different age groups. According to the physiological characteristics and radiation sensitivity research data of children, adolescents, adults and the elderly, an age weight factor is set in the dose control algorithm;

[0008] Image preprocessing module: connected to the image acquisition module, used to process the raw data to reduce image artifacts. The image preprocessing module adopts a metal artifact correction algorithm based on projection data preprocessing and a motion artifact correction algorithm based on image registration. The metal artifact correction algorithm performs secondary correction based on projection data preprocessing in combination with an iterative reconstruction algorithm. The metal area is first preliminarily identified and corrected through projection data preprocessing, and then the iterative reconstruction algorithm is used to further optimize the remaining artifacts during the reconstruction process, and the real image is gradually approached through multiple iterations. The motion artifact correction algorithm adds a real-time motion monitoring device to the scanning device, such as a high-precision acceleration sensor or an optical tracking unit. When a large-scale movement of the patient is detected, the scanning is immediately paused, and the patient is prompted to remain still and then rescan the area. At the same time, a motion trajectory and amplitude analysis module is added to the image registration algorithm to more accurately perform image registration and fusion according to the motion characteristics.

[0009] Image reconstruction module: connected to the image preprocessing module, used to reconstruct the processed image data into a three-dimensional image. The image reconstruction module adopts a super-resolution reconstruction algorithm to improve the spatial resolution of the image, and adds a noise suppression unit to the super-resolution reconstruction algorithm. The noise of the low-dose scanned image is first estimated, and then an adaptive filtering technology or a noise removal method based on deep learning is used to effectively suppress the noise while improving the image resolution. At the same time, a device adaptation library for the super-resolution reconstruction algorithm is established to collect parameter information and image feature data of CBCT devices of different manufacturers and models, and a specific algorithm adaptation module is developed for each device type. When in use, the CBCT device model is automatically identified and the corresponding adaptation module is called;

[0010] Data analysis module: connected to the image reconstruction module, the data analysis module includes an automatic bone facial type classification unit and an intelligent measurement unit. The automatic bone facial type classification unit automatically classifies the mandibular body morphology into high-angle type, low-angle type or average-angle type based on the angle, length and other parameters of the mandibular body. The unit continuously expands the training data set, including collecting more CBCT data of patients of different races, ages, genders and patients with various maxillofacial diseases, and performing various preprocessing and enhancement operations on the data, such as rotation, flipping, adding noise, etc., to improve the generalization ability of the algorithm and reduce the risk of overfitting and underfitting. At the same time, an integrated learning method is adopted to combine multiple different machine learning models for classification decisions; the intelligent measurement unit can automatically identify the key anatomical landmarks of the mandibular body, and automatically measure the morphological parameters based on these points. A multi-feature fusion and error correction mechanism is added to the automatic identification of anatomical landmarks. In addition to image feature recognition based on deep learning, the landmarks are located in combination with multi-dimensional information such as the geometric shape characteristics and spatial position relationship of the mandibular body. When there is a large deviation or uncertainty in the recognition result, a human-computer interaction method is adopted to prompt the doctor to make manual corrections;

[0011] Display and interaction module: connected to the data analysis module, used to display the analysis results and provide an interactive operation interface. The display and interaction module uses ray tracing and volume rendering technology to enhance the realism and stereoscopic sense of the image, and combines virtual reality or augmented reality technology to provide an immersive interactive experience. This module develops lightweight virtual reality or augmented reality applications based on smartphones or tablets, and uses the popularity and low-cost advantages of these mobile devices to transplant some interactive functions to mobile platforms. Through wireless connection to CBCT analysis equipment, basic image display and simple interactive operations such as zooming and rotating are achieved. In addition, a multi-person collaboration function module is added to the display and interaction module to establish a network-based image sharing platform, which supports multiple users to log in and operate the same set of mandibular body image data at the same time, and set different user authority levels, such as administrators, experts, observers, etc., to facilitate effective interactive collaboration during consultation or teaching, including real-time annotation, voice communication, synchronous operation, etc.

[0012] Furthermore, the image acquisition module includes an artificial intelligence-assisted scanning parameter optimization unit, which comprehensively considers factors such as the degree of abnormality in the appearance of the mandibular body, relevant disease information in the medical history, and the patient's body posture stability when generating initial scanning parameters, so as to further improve the accuracy of the initial scanning parameters.

[0013] Furthermore, during the iterative reconstruction process, the metal artifact correction algorithm in the image preprocessing module dynamically adjusts the number of iterations and correction parameters according to the intensity, distribution range and relative position relationship of the metal artifacts with the key structures of the mandibular body to achieve the best artifact correction effect.

[0014] Furthermore, the real-time motion monitoring device of the motion artifact correction algorithm in the image preprocessing module can distinguish between normal physiological movements of the patient, such as small displacements caused by heartbeat and breathing, and abnormal large-amplitude movements, and use a compensation algorithm to correct the image for normal physiological movements, thereby reducing unnecessary rescanning.

[0015] Furthermore, when the noise suppression unit of the super-resolution reconstruction algorithm in the image reconstruction module adopts the adaptive filtering technology, it adaptively adjusts the filtering parameters according to the noise characteristics of different areas of the image and the richness of the image details to avoid over-smoothing the image details.

[0016] Furthermore, when the automatic bone facial type classification unit in the data analysis module adopts the integrated learning method, it dynamically adjusts the weight of each model in the final decision according to the classification accuracy and misjudgment type of different machine learning models to improve the accuracy of classification.

[0017] Furthermore, the multi-feature fusion and error correction mechanism of the intelligent measurement unit in the data analysis module, when combined with the geometric shape features of the mandibular body, utilizes the symmetry, curvature change and other features of the mandibular body to assist in the positioning and correction of key anatomical landmarks.

[0018] Furthermore, the display and interaction module includes a lightweight virtual reality or augmented reality application that supports gesture recognition operations on a mobile device, making it easier for doctors to perform more natural and smooth image interaction when holding the device.

[0019] Furthermore, the device is used in the design of oral orthodontic treatment plans, oral and maxillofacial surgery planning, diagnosis of temporomandibular joint disorders, and oral medicine teaching. It provides a basis for clinical decision-making through accurate analysis of the mandibular morphology of different vertical facial types, and in multidisciplinary consultation scenarios, the multi-person collaboration function module based on the display and interaction module enables efficient collaborative analysis and discussion of mandibular images by experts from different departments.

[0020] Compared with the prior art, the present invention has the following beneficial effects:

[0021] 1. The metal artifact correction algorithm first uses projection data preprocessing to preliminarily locate the metal area, and then combines iterative reconstruction algorithm to dynamically optimize the number of iterations and correction parameters according to the metal artifact intensity, distribution range and relative position to the key structure, effectively removing metal artifact interference, making image details clearer and significantly improving image quality. The motion artifact correction algorithm uses a high-precision real-time motion monitoring device to accurately distinguish between physiological and abnormal movements. A compensation algorithm is used to correct the image for normal physiological movement to reduce unnecessary rescans; abnormal movement is paused and rescanned in a timely manner, and the motion analysis module in the image registration algorithm is used to accurately register and fuse images based on motion characteristics, thereby improving preprocessing efficiency and quality and reducing image errors;

[0022] 2. The image spatial resolution is greatly improved through the super-resolution reconstruction algorithm, making the fine structure of the mandible clearly discernible. Its noise suppression module first accurately evaluates the image noise level, and then adaptively adjusts the filter parameters according to the noise characteristics of different regions and the richness of image details, effectively reducing noise while retaining image details, enhancing image visual effects, and helping doctors make accurate diagnoses. The device adaptation library covers a variety of CBCT device parameters and image feature information, can automatically identify device models and call corresponding adaptation modules, to ensure that the super-resolution reconstruction algorithm runs stably and efficiently on different devices, enhance the system's versatility and compatibility, and meet the needs of different medical institutions;

[0023] 3. The automatic bone facial classification unit automatically classifies based on the multi-angle and length parameters of the mandible, continuously expands the multi-source data training set and performs diversified pre-processing enhancements, uses integrated learning combined with multi-model decision-making, and dynamically weights the model accuracy and misjudgment type to improve classification accuracy, providing a scientific basis for clinical treatment planning and disease diagnosis. The intelligent measurement unit integrates multiple features to automatically identify anatomical landmarks, combines geometric shapes with spatial position relationships to locate, and automatically triggers human-computer interaction when deviations occur, and doctors can manually correct them. The display and interaction module uses advanced technology to enhance the realism and three-dimensionality of the image, combines VR / AR to provide an immersive experience, develops mobile applications for convenient operation, and the multi-person collaboration function module supports multi-disciplinary consultation, collaborative analysis and discussion, promotes disciplinary exchanges and cooperation, and improves the overall diagnosis and treatment level and work efficiency of oral medicine. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 The present invention is a principle block diagram of a mandibular body morphology analysis device based on cone beam CT data in different vertical bone facial types and its application. DETAILED DESCRIPTION

[0025] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0026] See also Figure 1 The present invention provides a technical solution: a mandibular morphology analysis device based on cone beam CT data in different vertical bone facial types and its application, including assuming that a 35-year-old male patient comes to the oral orthodontics department for treatment due to uneven teeth and suspects that the mandibular morphology is abnormal, which may affect the orthodontic treatment plan. The doctor decides to use the mandibular morphology analysis device based on cone beam CT data to examine the patient.

[0027] Image acquisition phase:

[0028] After the patient entered the scanning room, the operator first examined the appearance of the patient's mandibular body and found that there was no obvious deformity, swelling or trauma to the mandibular body, but the patient had a history of mild temporomandibular joint clicking. At the same time, it was observed that the patient's body posture was stable, without obvious tremor or shaking.

[0029] The appearance of the mandibular body of the patient was recorded as 0 for no obvious abnormality and 1 for slight abnormality, and the medical history information, the clicking of the temporomandibular joint was recorded as a specific code, such as 101, and the patient's age, 35 years old, gender, male, and other information were input into the artificial intelligence-assisted scanning parameter optimization unit in the image acquisition module.

[0030] The unit automatically generates initial scan parameters based on a large amount of previous case data. For this patient, the generated field of view is 16cm×16cm, which is suitable for comprehensive observation of the mandibular body and surrounding structures. The tube current is set to 7mA and the tube voltage is 95kV. This is because according to the patient's age, appearance and medical history, this parameter combination can effectively control the radiation dose while ensuring that the image quality meets the analysis requirements.

[0031] Start the scan. During the scanning process, the image acquisition module uses variable field of view scanning technology to scan according to the preset 16cm×16cm field of view. Through a precise dose control algorithm, based on the patient's body shape, assuming a weight of 75kg and a height of 180cm, the radiation dose is automatically adjusted to ensure that the radiation dose is within a safe range and the image is clear.

[0032] Image preprocessing stage:

[0033] After the scan is completed, the raw data is transmitted to the image preprocessing module. This module first uses a metal artifact correction algorithm based on projection data preprocessing. First, the raw projection data transmitted to the image preprocessing module after the scan is completed is obtained. These data are a series of two-dimensional projection image sequences obtained by X-raying the area containing the mandibular body and possible foreign bodies such as metal dentures from different angles during the cone beam CT scan. Each raw projection image is analyzed, and a specific grayscale threshold is set to try to preliminarily identify possible metal areas. Generally speaking, metal usually presents a relatively high grayscale value in the projection image because its attenuation of X-rays is significantly different from that of the surrounding tissue. Therefore, a suitable grayscale threshold is determined based on experience and previous analysis of images containing metal samples. Then, each pixel in each projection image is traversed. If the grayscale value of the pixel is greater than the set threshold, it is marked as a suspected metal pixel, thereby preliminarily delineating the area where metal may exist. For the initially identified suspected metal area, further precise definition and feature extraction are performed. Through the corrosion operation, some isolated small pixel groups that may be misjudged can be removed to make the metal area more compact; while the expansion operation can restore the boundary information of the metal area that may be lost due to the corrosion operation to a certain extent, ensuring the integrity of the metal area. At the same time, some characteristic information of the metal area is extracted, such as the area size, shape (roughly circular, square or irregular shape, etc.) of the metal area, and the position coordinates in the projection image, etc., which will help to make up for the subsequent more accurate correction processing. According to the extracted metal area characteristics and the projection data information of the surrounding normal tissue, a suitable correction method is used to deal with metal artifacts. A common method is to use the projection data of the surrounding normal tissue to interpolate and replace the metal area. For example, a certain range of normal tissue areas are selected around the metal area, such as a ring area around the metal area, and the projection data characteristics of the normal tissue in the area are obtained. Then, based on these data, a certain interpolation algorithm, such as linear interpolation, spline interpolation, etc., is used to generate projection data to replace the metal area, thereby compensating for the accurate projection information missing due to the presence of metal to a certain extent and reducing the influence of metal artifacts on subsequent image reconstruction. After completing the above preliminary correction, a preliminary image reconstruction can be performed based on the corrected projection data to obtain a preliminary three-dimensional image. Then check this preliminary image to see if there are still obvious metal artifacts. If it is found that there are still obvious artifacts, it is necessary to adjust the previous correction steps according to the situation of the residual artifacts, such as re-adjusting the grayscale threshold, changing the parameters of the interpolation algorithm, or expanding / reducing the range of the surrounding normal tissue area, and then perform metal area identification, correction and image reconstruction again. Repeat this process until the impact of metal artifacts in the obtained image is reduced to an acceptable level, providing a relatively accurate projection data basis for subsequent image processing.

[0034] The motion artifact correction algorithm works at the same time, and the high-precision accelerometer on the scanning device monitors the patient's movement in real time. During the scan, it was detected that the patient had a brief movement of about 3mm due to swallowing. Due to the large amplitude of the movement, the motion artifact correction algorithm immediately paused the scan and prompted the patient to remain still and rescan the area. In the image registration algorithm, the added motion trajectory and amplitude analysis module records the characteristics of this movement, including the direction of movement, mainly vertical, the amplitude of movement 3mm, and the time point of movement, 10 seconds after the start of the scan. After rescanning, based on these motion characteristics, the data of the two scans are accurately registered through image registration and fusion algorithms to eliminate the influence of motion artifacts.

[0035] Image reconstruction stage:

[0036] The preprocessed image data is transmitted to the image reconstruction module. This module uses a super-resolution reconstruction algorithm to improve the spatial resolution of the image. In the super-resolution reconstruction algorithm, the noise level of the low-dose scanned image is first estimated. For example, the noise level is evaluated by calculating the variance of the grayscale values ​​in different areas of the image. It is found that the noise variance in the center area of ​​the image is 15, and the noise variance in the edge area is 25.

[0037] Then, the adaptive filtering technology is used. According to the small noise variance and rich image details in the central area of ​​the image, the number of edge pixels accounts for a high proportion of about 15%, and a smaller filter kernel 3x3 and a weaker filter strength are used; while in the edge area, due to the large noise variance and relatively few details, the number of edge pixels accounts for about 8%, and the filter kernel 5x5 and filter strength are appropriately increased. At the same time, the image reconstruction module automatically recognizes that the CBCT device model used is Smart3D-Xs-U400, and calls the corresponding algorithm adaptation module from the established super-resolution reconstruction algorithm device adaptation library to further optimize the reconstruction effect.

[0038] Data analysis phase:

[0039] The reconstructed three-dimensional image was input into the data analysis module. The automatic bone facial type classification unit first automatically classified the mandibular body morphology based on parameters such as the angle of the mandibular body, the mandibular plane angle was measured as 35°; the length, the mandibular body length was measured as 85mm. The unit continuously expanded the training data set and has collected CBCT data of 5,000 patients of different races, ages, genders and patients with various maxillofacial diseases. The data were preprocessed and enhanced in a variety of ways, such as randomly rotating the image ±5°, flipping the image horizontally or vertically, adding noise of different intensities, and the signal-to-noise ratio was 25-35dB. The ensemble learning method was used to combine three different machine learning models, support vector machine, decision tree and neural network, for classification decision making. In the decision-making process, according to the classification accuracy of different machine learning models, the support vector machine had an accuracy of 88%, the decision tree had an accuracy of 82%, and the neural network had an accuracy of 90%, and the misjudgment type, the support vector machine misjudged some low-angle types as equal-angle types, and dynamically adjusted the weights of each model in the final decision, and finally classified the mandibular body morphology of the patient as equal-angle type.

[0040] The intelligent measurement unit can automatically identify the key anatomical landmarks of the mandibular body, such as the mental foramen, mandibular angle, and condylar vertex. Multi-feature fusion and error correction mechanisms are added to the algorithm for automatic identification of anatomical landmarks. In addition to image feature recognition based on deep learning, the geometric shape characteristics of the mandibular body are combined with the symmetry of the mandibular body. By calculating the similarity index of the mandibular bodies on both sides, the SSIM value is obtained to be 0.82; the mandibular angle position is determined according to the change in curvature, and the curvature at the mandibular angle is 0.25; the spatial position relationship, the landmark range is determined according to the relative position relationship between the mandibular body and other skull structures, and the landmarks are located by multi-dimensional information. During the positioning process, it was found that the initial positioning error of the mental foramen was 2.5mm. Due to the large error, the human-computer interaction method was used to highlight the mental foramen position on the screen. The doctor manually corrected it by clicking the mouse to reduce the positioning error to less than 1mm. Then, based on these accurate landmarks, the morphological parameters are automatically measured, such as the width and height of the mandibular body, to provide data support for the subsequent orthodontic treatment plan design.

[0041] Display and interaction stage:

[0042] The data analysis results are transmitted to the display and interaction module. This module uses ray tracing and volume rendering technology to enhance the realism and stereoscopic sense of the image, so that the mandibular image can be displayed more clearly and realistically. Combined with virtual reality or augmented reality technology to provide an immersive interactive experience, doctors can wear a virtual reality helmet to observe the three-dimensional structure of the mandibular body in a virtual environment. They can use the handle to zoom in and out, such as from 1:1 to 1:5; rotate, such as rotating 30° around the X-axis and 45° around the Y-axis, etc., to more intuitively understand the morphological characteristics of the mandibular body.

[0043] At the same time, the developed lightweight virtual reality or augmented reality applications based on smartphones can also be used. Doctors use mobile phones to wirelessly connect to CBCT analysis equipment and realize basic image display and simple interactive operations on the mobile phone screen, such as pinch-to-zoom, sliding and rotating images, etc. The multi-person collaboration function module added to the display and interaction module plays a role. In the multidisciplinary consultation scenario, experts from different departments, such as orthodontists, oral and maxillofacial surgeons, and temporomandibular joint experts, can log in to the network-based image sharing platform at the same time. Orthodontists have full operation permissions as administrators, oral and maxillofacial surgeons have in-depth analysis and annotation permissions as experts, and temporomandibular joint experts have the permission to view images and discuss records as observers. Experts can annotate key parts on the image in real time, such as oral and maxillofacial surgeons annotating possible surgical areas of the mandibular body, and analyze the results, and discuss through voice communication functions, such as discussing the coordination between orthodontic treatment and possible surgical plans, and all operations are displayed synchronously on each client, realizing efficient collaborative analysis and discussion, and providing a basis for formulating a comprehensive treatment plan.

[0044] It should be noted here that: in the present invention, the scanning parameters lack precise personalization during image acquisition, and the differences in patient age, body shape, mandibular appearance and medical history are fully considered, so the image quality is high and the radiation hazard is reduced. In the image preprocessing stage, metal artifact correction is flexibly adjusted in combination with metal features, and motion artifacts are distinguished and compensated, which improves image quality and efficiency. During image reconstruction, the super-resolution algorithm properly handles noise and has good equipment adaptability. In the data analysis stage, the bone facial classification algorithm has strong generalization ability, the positioning error of the measurement unit landmark point is small, and the error correction mechanism is perfect. In terms of display and interaction, VR / AR applications are intuitively combined with mobile devices, and multidisciplinary consultations are efficient and collaborative.

Claims

1. A device for analyzing the mandibular body morphology in different vertical directions based on cone beam CT data, characterized in that: include: Image acquisition module: used to obtain cone beam CT raw data including the mandibular body, and equipped with a precise dose control algorithm, which can automatically adjust the radiation dose according to the patient's body shape and scanning site; Image preprocessing module: connected to the image acquisition module, used to process the raw data, the image preprocessing module adopts a metal artifact correction algorithm based on projection data preprocessing and a motion artifact correction algorithm based on image registration; wherein the metal artifact correction algorithm performs secondary correction based on projection data preprocessing combined with an iterative reconstruction algorithm; the motion artifact correction algorithm adds a real-time motion monitoring device to the scanning device; Image reconstruction module: connected to the image preprocessing module, used to reconstruct the processed image data into a three-dimensional image. The image reconstruction module adopts a super-resolution reconstruction algorithm to improve the image spatial resolution, and adds a noise suppression unit to the super-resolution reconstruction algorithm. It first estimates the noise of the low-dose scanned image, and then adopts adaptive filtering technology or a noise removal method based on deep learning to effectively suppress noise while improving the image resolution; Data analysis module: connected to the image reconstruction module, the data analysis module includes an automatic bone face classification unit and an intelligent measurement unit. The automatic bone face classification unit automatically classifies the mandibular body morphology into high-angle type, low-angle type or average-angle type based on the angle and length parameters of the mandibular body. The unit continuously expands the training data set; Display and interaction module: connected to the data analysis module, used to display the analysis results and provide an interactive operation interface.

2. The device for analyzing the mandibular body morphology in different vertical directions based on cone beam CT data according to claim 1, characterized in that: The image acquisition module includes an artificial intelligence-assisted scanning parameter optimization unit. When generating initial scanning parameters, it comprehensively considers the degree of abnormality in the appearance of the mandibular body, relevant disease information in the medical history, and the patient's body posture stability factors to further improve the accuracy of the initial scanning parameters.

3. The device for analyzing the mandibular body morphology in different vertical directions based on cone beam CT data as claimed in claim 1, characterized in that: The metal artifact correction algorithm in the image preprocessing module dynamically adjusts the number of iterations and correction parameters during the iterative reconstruction process according to the intensity and distribution range of the metal artifacts and their relative positional relationship with the key structures of the mandibular body.

4. The device for analyzing the mandibular body morphology in different vertical directions based on cone beam CT data according to claim 1, characterized in that: The real-time motion monitoring device of the motion artifact correction algorithm in the image preprocessing module can distinguish between normal physiological movements of the patient, such as small displacements caused by heartbeat and breathing, and abnormal large-amplitude movements.

5. The device for analyzing the mandibular body morphology in different vertical directions based on cone beam CT data as claimed in claim 1, characterized in that: When the noise suppression unit of the super-resolution reconstruction algorithm in the image reconstruction module adopts the adaptive filtering technology, it adaptively adjusts the filtering parameters according to the noise characteristics of different areas of the image and the richness of the image details to avoid over-smoothing the image details.

6. The device for analyzing the mandibular body morphology in different vertical directions based on cone beam CT data as claimed in claim 1, characterized in that: When the automatic bone facial type classification unit in the data analysis module adopts the integrated learning method, it dynamically adjusts the weight of each model in the final decision according to the classification accuracy and misjudgment type of different machine learning models to improve the accuracy of classification.

7. The device for analyzing the mandibular body morphology in different vertical directions based on cone beam CT data as claimed in claim 1, characterized in that: The multi-feature fusion and error correction mechanism of the intelligent measurement unit in the data analysis module, when combined with the geometric shape features of the mandibular body, utilizes the symmetry and curvature change characteristics of the mandibular body to assist in the positioning and correction of key anatomical landmarks.

8. The device for analyzing the mandibular body morphology in different vertical directions based on cone beam CT data as claimed in claim 1, characterized in that: The display and interaction module includes a lightweight virtual reality or augmented reality application that supports gesture recognition operations on a mobile device, making it easier for doctors to perform more natural and smooth image interaction when holding the device.

9. Application of a device for analyzing the mandibular body morphology in different vertical directions based on cone beam CT data according to any one of claims 1 to 8, characterized in that: The device is used in the design of orthodontic treatment plans, oral and maxillofacial surgery planning, diagnosis of temporomandibular joint disorders, and oral medicine teaching.

Images