A periodontal examination automatic auxiliary diagnosis method based on registration of oral cone beam CT image and oral scanning data

Abstract

The present application relates to the field of oral medical technology, in particular to a periodontal examination automatic auxiliary diagnosis method based on oral cone beam CT image and mouth scanning data registration, comprising collecting patient CBCT image and intraoral scanning data, analyzing CBCT image to extract tooth data body, tooth crown cut end feature, coordinates and periodontal information, identifying enamel and alveolar bone boundary; analyzing mouth scanning data to extract tooth crown cut end feature and gum line; integrating CBCT and mouth scanning data, calculating pocket depth, recession and attachment loss after registration through measuring points; displaying periodontal measurement results on the periodontal reading system and generating a report. The present application calculates the key points of periodontal examination, enamel boundary points, alveolar bone boundary points and gum line boundary points by respectively intersecting the boundary line with six measurement surfaces, and calculates pocket depth, recession and attachment loss, realizes automatic generation of periodontal examination results, visualizes and audits the condition of periodontal disease by combining CBCT reading tool, and is convenient for manual audit and doctor-patient communication.

Description

Technical Field

[0001] This invention relates to the field of oral medical technology, specifically to an automated assisted diagnostic method for periodontal examination based on registration of oral cone-beam CT images and intraoral scan data. Background Technology

[0002] Periodontal disease (primarily periodontitis) is a chronic inflammatory disease with insidious early symptoms, but its harm is far-reaching, not only damaging oral health but also potentially triggering systemic diseases. Regular professional periodontal checkups are one of the most effective ways to maintain healthy teeth and gums and prevent tooth loss. Good oral hygiene habits, regular professional checkups, and early, proactive intervention for periodontal disease are key to maintaining lifelong oral health. Once periodontal disease develops, professional periodontal treatment should be sought as soon as possible to control disease progression and understand possible future restorative options and their challenges.

[0003] Existing periodontal examination techniques are mainly divided into two types: artificial periodontal examination and periodontal imaging examination. Artificial periodontal examination relies heavily on the dentist's skill level, cannot accurately quantify bone loss, and is insensitive to buccal and lingual bone plate detection. Furthermore, it requires a series of pre-examination cleaning procedures, and probing is somewhat invasive, causing discomfort to the patient during the examination. Periodontal imaging examination relies on the dentist manually interpreting cone-beam computed tomography (CBCT) data and panoramic / periapical radiographs. Panoramic / periapical radiograph interpretation can only observe alveolar bone resorption at the mesial and distal points, and has low diagnostic specificity. CBCT data diagnosis can better cover the above problems, but it also suffers from inaccurate identification of gingival soft tissue recession, long interpretation time, and high subjectivity. Therefore, we propose an automated assisted diagnostic method for periodontal examination based on registration of oral cone-beam CT images and intraoral scan data. Summary of the Invention

[0004] In view of the above-mentioned shortcomings of the existing technology, the first objective of the present invention is to provide an automated auxiliary diagnostic method for periodontal examination based on the registration of oral cone-beam CT images and oral scan data, thereby solving the problems in the background technology.

[0005] To achieve the above objectives, the present invention provides the following technical solution: An automated assisted diagnostic method for periodontal examination based on registration of oral cone-beam CT images and intraoral scan data includes the following steps: S1. Acquire CBCT images and intraoral scan data of the patient, or directly use pre-acquired CBCT images and intraoral scan data; S2. Analyze CBCT images to extract tooth data volumes, crown incisal feature points, tooth coordinate orientation information, and periodontal measurement surface information, and identify the enamel boundary and alveolar bone boundary from the tooth data volumes; S3. Analyze the patient's oral scan data to identify the characteristic points of the incisal edge of the crown and the measurement data of the tooth-gingival division line; S4. Summarize the calculation results of S2 and S3. By registering the tooth data of the oral scan model with the tooth data extracted from CBCT, the enamel boundary, alveolar bone boundary and tooth-gingival separation line are aligned to the unified world coordinate system. Three measurement points are obtained in each of the six measurement planes. The height difference along the long axis of the tooth is calculated by simulating the periodontal probe to obtain the measurement results of pocket depth, retraction and attachment loss. S5. Display the periodontal measurement results for each tooth position in the periodontal radiograph reading system and generate a periodontal examination report; In step S5, the periodontal measurement results of each tooth position are displayed in the periodontal radiograph reading system, and a periodontal examination report is generated. The pocket depth, recession, and attachment loss information of each tooth calculated in S4 are displayed in the radiograph reading module M05. The six periodontal probing sites of each tooth position are numerically displayed and interface prompts are provided. When the attachment loss (CAL) is >4mm, a warning message will be displayed. When the warning message of the corresponding tooth position is clicked, the position of the corresponding measurement surface is quickly located and displayed in the MPR view so that the doctor can manually review it and communicate with the doctor and patient. Three measurement points were obtained from each of the six measurement planes. The height difference along the long axis of the tooth at these three points was calculated using a simulated periodontal probe to obtain the pocket depth, recession, and attachment loss measurements. In the periodontal examination analysis module M04, the CT tooth segmentation data, tooth long axis, mesiodistal axis, buccal-lingual axis, crown incisal feature points, enamel boundary line, alveolar bone boundary line, and six measurement planes calculated in S2, along with the tooth segmentation data, tooth-gingival segmentation line, and crown incisal feature points calculated from the oral scan data, were integrated. The CBCT data and oral scan data were then registered, and the periodontal diagnostic measurement data of the teeth were calculated. The crown incisal feature points identified in the CBCT data were compared with those in the oral scan data. The incisal feature points of the crowns identified in the data have spatial similarity. Based on this information, the iterative nearest point algorithm is used to register the 16 incisal feature points of the crowns extracted from the CBCT data with the 16 incisal feature points of the crowns identified from the intraoral scan data to obtain the registration matrix of the maxilla. The same registration matrix of the mandible is obtained. The intraoral scan model and CT data are registered to a unified spatial coordinate system, and the accuracy of the registration is observed. The registration matrix is ​​used to map the tooth-gingival segmentation data extracted from the intraoral scan data to the unified spatial coordinate system of the CBCT data. For each tooth, three measurement curves in the same coordinate system can be obtained: enamel boundary line, alveolar bone boundary line, and tooth-gingival segmentation line. The three measurement curves mentioned above are passed through the six measurement planes calculated in S3. Each plane will have three intersection points: the enamel boundary point, the alveolar bone boundary point, and the gingival margin boundary point. Taking the tooth center as the origin and the tooth's long axis as the Y-axis, the heights of the three feature points along the Y-axis are calculated and denoted as Hcej, Habm, and Hgm, respectively. Then: PD = Hgm - Habm, REC=Hcej - Hgm, CAL = Hcej - Habm; This yields the calculated pocket depth (PD), retraction (REC), and attachment loss (CAL) for each tooth.

[0006] The present invention is further configured such that: in step S1, the patient's CBCT images and intraoral scan data are acquired, or the patient's oral cavity image data is acquired through the CBCT device in the M01 image acquisition module, and the data is stored and transmitted in the standard DICOM method, and CBCT data input from the outside is also received.

[0007] The present invention is further configured such that: in step S1, the patient's CBCT images and intraoral scan data are acquired, or the patient's oral cavity image data is acquired through the CBCT device in the M01 image acquisition module, and the data is stored and transmitted in the standard DICOM method, and CBCT data input from the outside is also received.

[0008] The present invention is further configured as follows: In step S2.1, the CBCT data of S1 is input into the M02 CBCT image analysis module for analysis, and the three-dimensional segmentation information and feature information of the teeth and jawbone are extracted respectively. After reading the image DICOM data, the CBCT is reconstructed by a filtering back projection algorithm, and an artificial intelligence algorithm based on a three-dimensional U-shaped convolutional neural network is used to identify and extract the segmentation data of the teeth and jawbone in the CT data, and output the health status of each tooth. The health status of the teeth has important reference value for the diagnosis of periodontal disease. Next, a three-dimensional convolutional neural network is used to identify the crown center O, tooth direction and crown incisal feature point I of the extracted tooth segmentation data.

[0009] The present invention is further configured as follows: In step S2.2, after obtaining all the above calculation results, the tooth data information and spatial coordinate information are used to calculate the information of each measurement surface of the tooth, namely, the six measurement surfaces: mesobalvular plane Me, centric buccal plane M, distobuccal plane D, mesobalvular plane Me1, centric lingual plane M1, and distolingual plane D1. Among these, the centric buccal plane M and the centric lingual plane M1 are jointly determined by the buccal-lingual axis and the long axis of the tooth, with M on the buccal side and M1 on the lingual side. The intersection point of the buccal half-axis of the buccal-lingual axis with the tooth surface is denoted as PtM, and the intersection point of the lingual half-axis with the tooth surface is denoted as PtM1. The mesobal and distal measurement planes need to consider that the probe cannot reach the adjacent areas of the mesobal and distal sides of the two teeth, and the interproximal spacing is set to... Set to 1mm, iterate through the minimum distances from all points at the tooth's center height to the mesial and distal surfaces of the two adjacent teeth. When the minimum distance is less than 1mm, mark it as the proximal region. Select the point in the mesial region closest to PtM as the mesial buccal point PtMe, and the point in the distal region closest to PtM as the distal buccal point PtD. Calculate the mesial lingual point PtMe1 and the distal lingual point PtD1 on the lingual side. PtMe and the tooth's long axis determine the mesial buccal plane Me, PtD and the tooth's long axis determine the distal buccal plane D, PtMe1 and the tooth's long axis determine the mesial lingual plane Me1, and PtD1 and the tooth's long axis determine the mesial lingual plane D1. At this point, all six measurement points and planes have been determined.

[0010] The present invention is further configured as follows: In step S2.3, the three-dimensional data volume of teeth and jawbone extracted by the above CT is further processed to extract the enamel boundary line and alveolar bone boundary line. There are significant gray-level differences between enamel and bone, and between alveolar bone and teeth in the CBCT data. The gray-level differences are used to perform algorithmic analysis on the tooth data and jawbone data respectively to obtain the enamel boundary line and alveolar bone boundary line. The method used here is the three-dimensional watershed algorithm. First, the pixels and their gray-level values ​​of the three-dimensional segmentation data of the teeth are extracted. The gradient difference is calculated by the three-dimensional Sobel operator to obtain the gradient map. The center C of the tooth and the center point of the upper half of the structure on the horizontal plane are set as the crown seed point. The center C of the tooth and the center of the lower half of the horizontal plane are set as the root seed point. The segmentation boundary surface of the crown and tooth, i.e. the enamel boundary, is calculated according to the gradient map. The boundary condition of the enamel boundary within a certain horizontal thickness on the tooth surface is calculated. The closed curve formed by the lowest point along the long axis of the tooth within the thickness is used as the enamel boundary line. At the same time, the intersection operation of the jawbone segmentation result and the tooth segmentation result is performed. After subtracting the tooth data volume, a three-dimensional data volume containing only the jawbone is obtained. The highest point of the jawbone data volume along the long axis of the tooth within a certain horizontal thickness on the tooth surface is the alveolar bone boundary. The boundary is connected into a closed curve and smoothed to obtain the alveolar bone boundary line. The patient's oral scan data was analyzed to identify incisal feature points of the crown, tooth-gingival separation lines, and tooth segmentation measurement data. The patient's oral scan data was input into the oral scan data analysis module M03 for AI analysis and segmentation to extract three-dimensional data of the crown and gingiva. Incisal feature points of the crown were identified. A convolutional neural network algorithm based on PointNet and MeshSegNet models was used to identify and segment teeth in the oral scan model, obtaining tooth segmentation data and tooth-gingival separation line data for each tooth position. The segmented tooth model was further analyzed, and a convolutional neural network algorithm based on the PointNet model was used to identify the long axis of the teeth and incisal feature points of the crown.

[0011] The present invention is further configured such that: in step S4, the calculation results of S2 and S3 are summarized, and the tooth data of the oral scanning model are registered with the tooth data extracted from CBCT, and the enamel boundary, alveolar bone boundary and tooth-gingival separation line are aligned to a unified world coordinate system.

[0012] Beneficial effects Compared with known public technologies, the technical solution provided by this invention has the following beneficial effects: This invention utilizes AI-automated algorithms to analyze CBCT images and intraoral scan data, identifying and extracting characteristic information of teeth and jawbones. It registers CBCT image data and intraoral scan data using tooth feature points, unifying data and measurement results from different models into a single spatial coordinate system. The invention extracts the enamel boundary line, alveolar bone boundary line, and tooth-gingival separation line. By intersecting these boundary lines with six measurement planes, it calculates key points for periodontal examination: the enamel boundary point (CEJ), alveolar bone boundary point (ABM), and gingival margin boundary point (GM). It also calculates pocket depth (PD), recession (REC), and attachment loss (CAL), achieving fully automated generation of periodontal examination results. Combined with CBCT image reading tools, it provides a visual display and review of periodontal disease conditions, facilitating manual review and doctor-patient communication. Attached Figure Description

[0013] Figure 1 This is a schematic diagram of the system operation flow of an automated auxiliary diagnostic method for periodontal examination based on registration of oral cone-beam CT images and intraoral scan data according to the present invention. Figure 2 This is a schematic diagram illustrating the steps of an automated auxiliary diagnostic method for periodontal examination based on registration of oral cone-beam CT images and intraoral scan data according to the present invention. Figure 3 This is a schematic diagram of six measurement points for an automated auxiliary diagnostic method for periodontal examination based on registration of oral cone-beam CT images and intraoral scan data according to the present invention. Figure 4This is a schematic diagram of CBCT data and oral scan data feature registration for an automated auxiliary diagnostic method for periodontal examination based on the registration of oral cone-beam CT images and oral scan data according to the present invention. Figure 5 This is a schematic diagram illustrating the measurement principle of pocket depth, retraction, and attachment loss in an automated auxiliary diagnostic method for periodontal examination based on registration of oral cone-beam CT images and oral scan data according to the present invention. Figure 6 This is a schematic diagram illustrating the measurement of pocket depth, retraction, and attachment loss in an automated auxiliary diagnostic method for periodontal examination based on registration of oral cone-beam CT images and intraoral scan data according to the present invention. Figure 7 This is a schematic diagram of periodontal examination results from an automated assisted diagnostic method for periodontal examination based on registration of oral cone-beam CT images and intraoral scan data, as described in this invention. Figure 1 ; Figure 8 This is a schematic diagram of periodontal examination results from an automated assisted diagnostic method for periodontal examination based on registration of oral cone-beam CT images and intraoral scan data, as described in this invention. Figure 2 . Detailed Implementation

[0014] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0015] The present invention will be further described below with reference to embodiments.

[0016] Example 1 like Figure 1-8 As shown, the present invention provides a technical solution: an automated assisted diagnostic method for periodontal examination based on registration of oral cone-beam CT images and intraoral scan data, comprising the following steps: S1. Acquire CBCT images and intraoral scan data of the patient, or directly use pre-acquired CBCT images and intraoral scan data; S2. Analyze CBCT images to extract tooth data volumes, crown incisal feature points, tooth coordinate orientation information, and periodontal measurement surface information, and identify the enamel boundary and alveolar bone boundary from the tooth data volumes; S3. Analyze the patient's oral scan data to identify the characteristic points of the incisal edge of the crown and the measurement data of the tooth-gingival division line; S4. Summarize the calculation results of S2 and S3. By registering the tooth data of the oral scan model with the tooth data extracted from CBCT, the enamel boundary, alveolar bone boundary and tooth-gingival separation line are aligned to the unified world coordinate system. Three measurement points are obtained in each of the six measurement planes. The height difference along the long axis of the tooth is calculated by simulating the periodontal probe to obtain the measurement results of pocket depth, retraction and attachment loss. S5. Display the periodontal measurement results for each tooth position in the periodontal radiograph reading system and generate a periodontal examination report; In step S5, the periodontal measurement results of each tooth position are displayed in the periodontal radiograph reading system, and a periodontal examination report is generated. The pocket depth, recession, and attachment loss information of each tooth calculated in S4 are displayed in the radiograph reading module M05. The six periodontal probing sites of each tooth position are numerically displayed and interface prompts are provided. When the attachment loss (CAL) is >4mm, a warning message will be displayed. When the warning message of the corresponding tooth position is clicked, the position of the corresponding measurement surface is quickly located and displayed in the MPR view so that the doctor can manually review it and communicate with the doctor and patient. Three measurement points were obtained from each of the six measurement planes. The height difference along the long axis of the tooth at these three points was calculated using a simulated periodontal probe to obtain the pocket depth, recession, and attachment loss measurements. In the periodontal examination analysis module M04, the CT tooth segmentation data, tooth long axis, mesiodistal axis, buccal-lingual axis, crown incisal feature points, enamel boundary line, alveolar bone boundary line, and six measurement planes calculated in S2, along with the tooth segmentation data, tooth-gingival segmentation line, and crown incisal feature points calculated from the oral scan data, were integrated. The CBCT data and oral scan data were then registered, and the periodontal diagnostic measurement data of the teeth were calculated. The crown incisal feature points identified in the CBCT data were compared with those in the oral scan data. The incisal feature points of the crowns identified in the data have spatial similarity. Based on this information, the iterative nearest point algorithm is used to register the 16 incisal feature points of the crowns extracted from the CBCT data with the 16 incisal feature points of the crowns identified from the intraoral scan data to obtain the registration matrix of the maxilla. The same registration matrix of the mandible is obtained. The intraoral scan model and CT data are registered to a unified spatial coordinate system, and the accuracy of the registration is observed. The registration matrix is ​​used to map the tooth-gingival segmentation data extracted from the intraoral scan data to the unified spatial coordinate system of the CBCT data. For each tooth, three measurement curves in the same coordinate system can be obtained: enamel boundary line, alveolar bone boundary line, and tooth-gingival segmentation line. The three measurement curves mentioned above are passed through the six measurement planes calculated in S3. Each plane will have three intersection points: the enamel boundary point, the alveolar bone boundary point, and the gingival margin boundary point. Taking the tooth center as the origin and the tooth's long axis as the Y-axis, the heights of the three feature points along the Y-axis are calculated and denoted as Hcej, Habm, and Hgm, respectively. Then: PD = Hgm - Habm, REC=Hcej - Hgm, CAL = Hcej - Habm; This yields the calculated pocket depth (PD), retraction (REC), and attachment loss (CAL) for each tooth; In step S1, the patient's CBCT images and intraoral scan data are acquired, or the patient's oral cavity image data is acquired through the CBCT device in the M01 image acquisition module, using the pre-acquired CBCT images and intraoral scan data. The data is stored and transmitted in the standard DICOM format, and CBCT data input from external sources is also received. In step S1, the patient's CBCT images and intraoral scan data are acquired, or the patient's oral cavity image data is acquired through the CBCT device in the M01 image acquisition module, using the pre-acquired CBCT images and intraoral scan data. The data is stored and transmitted in the standard DICOM format, and CBCT data input from external sources is also received. In step S2.1, the CBCT data from S1 is input into the M02 CBCT image analysis module for analysis. The three-dimensional segmentation information and feature information of the teeth and jawbone are extracted. After reading the image DICOM data, the CBCT is reconstructed through a filtering back projection algorithm. An artificial intelligence algorithm based on a three-dimensional U-shaped convolutional neural network is used to identify and extract the segmentation data of the teeth and jawbone in the CT data, and output the health status of each tooth. The health status of the teeth has important reference value for the diagnosis of periodontal disease. Next, a three-dimensional convolutional neural network is used to identify the crown center O, tooth direction, and crown incisal feature point I of the extracted tooth segmentation data. In step S2.2, after obtaining all the above calculation results, the tooth data information and spatial coordinate information are used to calculate the information of each measurement plane of the tooth, namely the six measurement planes: mesobuccal plane Me, centric buccal plane M, distobuccal plane D, mesobuccal plane Me1, centric lingual plane M1, and distolingual plane D1. Among them, the centric buccal plane M and the centric lingual plane M1 are jointly determined by the buccal-lingual axis and the long axis of the tooth. M is on the buccal side and M1 is on the lingual side. The intersection point of the buccal half-axis of the buccal-lingual axis with the tooth surface is recorded as PtM, and the intersection point of the lingual half-axis with the tooth surface is recorded as PtM1. For the mesobuccal and distal measurement planes, it is necessary to consider that the probe cannot reach the adjacent areas of the mesobuccal and distal sides of the two teeth, and the interproximal spacing is set to 1mm. The algorithm iterates through all points at the tooth's center height and calculates the minimum distance from the mesial and distal surfaces of the adjacent teeth. When the minimum distance is less than 1 mm, it is marked as a proximal region. The point in the mesial region closest to PtM is selected as the mesial buccal point PtMe, and the point in the distal region closest to PtM is selected as the distal buccal point PtD. The mesial lingual point PtMe1 and the distal lingual point PtD1 are calculated on the lingual side. PtMe and the tooth's long axis determine the mesial buccal plane Me, PtD and the tooth's long axis determine the distal buccal plane D, PtMe1 and the tooth's long axis determine the mesial lingual plane Me1, and PtD1 and the tooth's long axis determine the mesial lingual plane D1. At this point, all six measurement points and planes have been determined. S2.3 further processes the three-dimensional data of teeth and jawbone extracted from the CT scan, extracting the enamel boundary and alveolar bone boundary. Significant grayscale differences exist between enamel and bone, and between alveolar bone and teeth in the CBCT data. These grayscale differences are used to analyze the tooth and jawbone data using algorithms to obtain the enamel boundary and alveolar bone boundary. The method used here is the three-dimensional watershed algorithm. First, the pixels and their grayscale values ​​of the three-dimensional segmented tooth data are extracted. The gradient difference is calculated using the three-dimensional Sobel operator to obtain the gradient map. The tooth center C and the center point of the upper half of the structure on the horizontal plane are set as the crown seed point. C and the center of the lower half of the horizontal plane are set as the root seed point. The segmentation boundary surface of the crown and tooth, i.e. the enamel boundary, is calculated according to the gradient map. The boundary condition of the enamel boundary within a certain horizontal thickness on the tooth surface is calculated. The closed curve formed by the lowest point along the long axis of the tooth within the thickness is used as the enamel boundary line. At the same time, the intersection operation of the jawbone segmentation result and the tooth segmentation result is performed. After subtracting the tooth data volume, a three-dimensional data volume containing only the jawbone is obtained. The highest point of the jawbone data volume along the long axis of the tooth within a certain horizontal thickness on the tooth surface is the alveolar bone boundary. The boundary is connected into a closed curve and smoothed to obtain the alveolar bone boundary line. The patient's oral scan data was analyzed to identify the incisal feature points of the crown, the tooth-gingival separation line, and the tooth segmentation data measurement data. The patient's oral scan data was input into the oral scan data analysis module M03 for AI analysis and segmentation to extract the three-dimensional data of the crown and gingiva. The incisal feature points of the crown were identified. The convolutional neural network algorithm based on the PointNet and MeshSegNet models was used to identify and segment the teeth in the oral scan model. The segmentation data of each tooth position and the tooth-gingival separation line data were obtained. The segmented tooth model was further analyzed, and the convolutional neural network algorithm based on the PointNet model was used to identify its tooth long axis and incisal feature points of the crown. In step S4, the calculation results of S2 and S3 are summarized. By registering the tooth data of the oral scan model with the tooth data extracted from CBCT, the enamel boundary, alveolar bone boundary and tooth-gingival separation line are aligned to a unified world coordinate system.

[0017] In this embodiment, through comprehensive analysis and registration of CBCT images and intraoral scan data, a precise assessment of the patient's periodontal health status is achieved. Three-dimensional segmentation technology is used to extract characteristic data of teeth and alveolar bone, identify characteristic points of the crown incisal edge and the tooth-gingival separation line, and spatially register the two to generate measurement data in a unified coordinate system. By accurately calibrating the enamel boundary, alveolar bone boundary, and tooth-gingival separation line, periodontal diagnostic indicators such as pocket depth (PD), recession (REC), and attachment loss (CAL) are calculated, greatly improving the accuracy and automation of diagnosis. The system displays a detailed periodontal examination report, assisting doctors in timely detection and treatment of periodontal problems, thereby improving diagnostic efficiency, reducing human error, and optimizing the patient's treatment process.

[0018] Working principle: like Figure 1-8 As shown, by acquiring the patient's CBCT images and intraoral scan data, or by using pre-acquired data for processing, these data are input into the system for analysis. The CBCT images are processed by a specialized module to extract three-dimensional segmentation data of the teeth and jawbone, including the boundary between the enamel and alveolar bone. The intraoral scan data is processed to identify key data such as the incisal feature points of the crown and the tooth-gingival separation line. Through in-depth analysis of these data, the system can perform a comprehensive periodontal examination of each tooth position on six measurement planes, including the precise positioning of the enamel boundary, alveolar bone boundary, and tooth-gingival separation line.

[0019] After processing and analyzing the data, the system spatially registers the intraoral scan data and CBCT data. This process ensures that the feature points in the two data sources are accurately aligned in the same coordinate system by matching the feature points. Through this registration, the system can obtain important periodontal parameters such as pocket depth (PD), recession (REC), and attachment loss (CAL) for each tooth. These parameters are obtained by simulating the measurement of the periodontal probe along the long axis of the tooth, providing accurate quantitative basis for the diagnosis of periodontal disease. By displaying these measurement results, the system helps doctors to more intuitively understand the periodontal health status of each tooth.

[0020] In practical use, the automation of the entire process greatly improves the efficiency and accuracy of periodontal examinations. Doctors only need to check the periodontal health status of each tooth through the reports provided by the system. When the system detects attachment loss greater than 4mm, it will also issue a warning to help doctors discover and deal with potential periodontal problems in a timely manner. Through this automated assisted diagnostic method, human error can be reduced, more accurate treatment suggestions can be provided, and the patient's treatment process can be accelerated.

[0021] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions will not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An automated assisted diagnostic method for periodontal examination based on registration of oral cone-beam CT images and intraoral scan data, characterized in that, Includes the following steps: S1. Acquire CBCT images and intraoral scan data of the patient, or directly use pre-acquired CBCT images and intraoral scan data; S2. Analyze CBCT images to extract tooth data volumes, crown incisal feature points, tooth coordinate orientation information, and periodontal measurement surface information, and identify the enamel boundary and alveolar bone boundary from the tooth data volumes; S3. Analyze the patient's oral scan data to identify the characteristic points of the incisal edge of the crown and the measurement data of the tooth-gingival division line; S4. Summarize the calculation results of S2 and S3. By registering the tooth data of the oral scan model with the tooth data extracted from CBCT, the enamel boundary, alveolar bone boundary and tooth-gingival separation line are aligned to the unified world coordinate system. Three measurement points are obtained in each of the six measurement planes. The height difference along the long axis of the tooth is calculated by simulating the periodontal probe to obtain the measurement results of pocket depth, retraction and attachment loss. S5. Display the periodontal measurement results for each tooth position in the periodontal radiograph reading system and generate a periodontal examination report; In step S5, the periodontal measurement results of each tooth position are displayed in the periodontal radiograph reading system, and a periodontal examination report is generated. The pocket depth, recession, and attachment loss information of each tooth calculated in S4 are displayed in the radiograph reading module M05. The six periodontal probing sites of each tooth position are numerically displayed and interface prompts are provided. When the attachment loss (CAL) is >4mm, a warning message will be displayed. When the warning message of the corresponding tooth position is clicked, the position of the corresponding measurement surface is quickly located and displayed in the MPR view so that the doctor can manually review it and communicate with the doctor and patient. Three measurement points were obtained from each of the six measurement planes. The height difference along the long axis of the tooth at these three points was calculated using a simulated periodontal probe to obtain the pocket depth, recession, and attachment loss measurements. In the periodontal examination analysis module M04, the CT tooth segmentation data, tooth long axis, mesiodistal axis, buccal-lingual axis, crown incisal feature points, enamel boundary line, alveolar bone boundary line, and six measurement planes calculated in S2, along with the tooth segmentation data, tooth-gingival segmentation line, and crown incisal feature points calculated from the oral scan data, were integrated. The CBCT data and oral scan data were then registered, and the periodontal diagnostic measurement data of the teeth were calculated. The crown incisal feature points identified in the CBCT data were compared with those in the oral scan data. The incisal feature points of the crowns identified in the data have spatial similarity. Based on this information, the iterative nearest point algorithm is used to register the 16 incisal feature points of the crowns extracted from the CBCT data with the 16 incisal feature points of the crowns identified from the intraoral scan data to obtain the registration matrix of the maxilla. The same registration matrix of the mandible is obtained. The intraoral scan model and CT data are registered to a unified spatial coordinate system, and the accuracy of the registration is observed. The registration matrix is ​​used to map the tooth-gingival segmentation data extracted from the intraoral scan data to the unified spatial coordinate system of the CBCT data. For each tooth, three measurement curves in the same coordinate system can be obtained: enamel boundary line, alveolar bone boundary line, and tooth-gingival segmentation line. The three measurement curves mentioned above are passed through the six measurement planes calculated in S3. Each plane will have three intersection points: the enamel boundary point, the alveolar bone boundary point, and the gingival margin boundary point. Taking the tooth center as the origin and the tooth's long axis as the Y-axis, the heights of the three feature points along the Y-axis are calculated and denoted as Hcej, Habm, and Hgm, respectively. Then: PD = Hgm - Habm, REC=Hcej - Hgm, CAL = Hcej - Habm; This yields the calculated pocket depth (PD), retraction (REC), and attachment loss (CAL) for each tooth.

2. The automated auxiliary diagnostic method for periodontal examination based on registration of oral cone-beam CT images and intraoral scan data as described in claim 1, characterized in that: Step S1 involves acquiring the patient's CBCT images and intraoral scan data, or directly using pre-acquired CBCT images and intraoral scan data. The CBCT device in the M01 image acquisition module acquires the patient's oral cavity image data. The data is stored and transmitted in the standard DICOM format, and CBCT data input from external sources is also received.

3. The automated auxiliary diagnostic method for periodontal examination based on registration of oral cone-beam CT images and intraoral scan data as described in claim 1, characterized in that: Step S2 involves analyzing the CBCT images to extract tooth data volumes, incisal feature points of the crown, tooth coordinate orientation information, and periodontal measurement surface information. From the tooth data volumes, the enamel boundary and alveolar bone boundary are identified. S2.1 Input the CBCT data from S1 into the M02 CBCT image analysis module for analysis, and extract the three-dimensional segmentation information and feature information of the teeth and jawbone respectively; S2.2 After obtaining all the above calculation results, use the tooth data information and spatial coordinate information to calculate the information of each measurement surface of the tooth, namely the six measurement surfaces: mesobuccal plane Me, centric buccal plane M, distobuccal plane D, mesobuccal plane Me1, centric lingual plane M1, and distolingual plane D1. S2.3 Further process the three-dimensional data of teeth and jawbone extracted from the above CT scan to extract the enamel boundary and alveolar bone boundary.

4. The automated assisted diagnostic method for periodontal examination based on registration of oral cone-beam CT images and intraoral scan data as described in claim 3, characterized in that: In step S2.1, the CBCT data from S1 is input into the M02 CBCT image analysis module for analysis. The three-dimensional segmentation information and feature information of the teeth and jawbone are extracted. After reading the image DICOM data, the CBCT is reconstructed using a filtered back projection algorithm. An artificial intelligence algorithm based on a three-dimensional U-shaped convolutional neural network is used to identify and extract the segmentation data of the teeth and jawbone from the CT data, and output the health status of each tooth. The health status of the teeth has important reference value for the diagnosis of periodontal disease. Next, a three-dimensional convolutional neural network is used to identify the crown center O, tooth direction, and crown incisal feature point I of the extracted tooth segmentation data.

5. The automated assisted diagnostic method for periodontal examination based on registration of oral cone-beam CT images and intraoral scan data as described in claim 3, characterized in that: In step S2.2, after obtaining all the above calculation results, the tooth data information and spatial coordinate information are used to calculate the information of each measurement plane of the tooth, namely the six measurement planes: mesobuccal plane Me, centric buccal plane M, distobuccal plane D, mesobuccal plane Me1, centric lingual plane M1, and distolingual plane D1. Among them, the centric buccal plane M and the centric lingual plane M1 are jointly determined by the buccal-lingual axis and the long axis of the tooth. M is on the buccal side and M1 is on the lingual side. The intersection point of the buccal half-axis of the buccal-lingual axis with the tooth surface is recorded as PtM, and the intersection point of the lingual half-axis with the tooth surface is recorded as PtM1. For the mesobuccal and distal measurement planes, it is necessary to consider that the probe cannot reach the adjacent areas of the mesobuccal and distal sides of the two teeth, and the interproximal spacing is set to 1mm. The algorithm iterates through all points at the tooth's center height and calculates the minimum distance from the points on the mesial and distal surfaces of the adjacent teeth. When the minimum distance is less than 1 mm, it is marked as a proximal region. The point in the mesial region closest to PtM is selected as the mesial buccal point PtMe, and the point in the distal region closest to PtM is selected as the distal buccal point PtD. The mesial lingual point PtMe1 and the distal lingual point PtD1 are calculated on the lingual side. PtMe and the tooth's long axis determine the mesial buccal plane Me, PtD and the tooth's long axis determine the distal buccal plane D, PtMe1 and the tooth's long axis determine the mesial lingual plane Me1, and PtD1 and the tooth's long axis determine the mesial lingual plane D1. At this point, all six measurement points and planes have been determined.

6. The automated assisted diagnostic method for periodontal examination based on registration of oral cone-beam CT images and intraoral scan data as described in claim 3, characterized in that: S2.3 further processes the three-dimensional data of teeth and jawbone extracted from the CT scan, extracting the enamel boundary and alveolar bone boundary. Significant grayscale differences exist between enamel and bone, and between alveolar bone and teeth in the CBCT data. These grayscale differences are used to analyze the tooth and jawbone data using algorithms to obtain the enamel boundary and alveolar bone boundary. The method used here is the three-dimensional watershed algorithm. First, the pixels and their grayscale values ​​of the three-dimensional segmented tooth data are extracted. The gradient difference is calculated using the three-dimensional Sobel operator to obtain the gradient map. The tooth center C and the center point of the upper half of the structure on the horizontal plane are set as the crown seed point. C and the center of the lower half of the horizontal plane are set as the root seed point. The segmentation boundary surface of the crown and tooth, i.e. the enamel boundary, is calculated according to the gradient map. The boundary condition of the enamel boundary within a certain horizontal thickness on the tooth surface is calculated. The closed curve formed by the lowest point along the long axis of the tooth within the thickness is used as the enamel boundary line. At the same time, the intersection operation of the jawbone segmentation result and the tooth segmentation result is performed. After subtracting the tooth data volume, a three-dimensional data volume containing only the jawbone is obtained. The highest point of the jawbone data volume along the long axis of the tooth within a certain horizontal thickness on the tooth surface is the alveolar bone boundary. The boundary is connected into a closed curve and smoothed to obtain the alveolar bone boundary line. The patient's oral scan data was analyzed to identify incisal feature points of the crown, tooth-gingival separation lines, and tooth segmentation measurement data. The patient's oral scan data was input into the oral scan data analysis module M03 for AI analysis and segmentation to extract three-dimensional data of the crown and gingiva. Incisal feature points of the crown were identified. A convolutional neural network algorithm based on PointNet and MeshSegNet models was used to identify and segment teeth in the oral scan model, obtaining tooth segmentation data and tooth-gingival separation line data for each tooth position. The segmented tooth model was further analyzed, and a convolutional neural network algorithm based on the PointNet model was used to identify the long axis of the teeth and incisal feature points of the crown.

Images