Temporomandibular joint three-dimensional shape measuring method and device and medium

Abstract

The invention discloses a temporomandibular joint three-dimensional shape measuring method and device and a medium, and relates to the field of image processing, and the method comprises the steps: inputting target temporomandibular joint CBCT data into a mandible three-dimensional segmentation model to obtain a mandible three-dimensional segmentation result; based on the mandible three-dimensional segmentation result and the target temporomandibular joint CBCT data, determining a temporomandibular joint central position area image; inputting the temporal-mandibular joint central position area image into the condyle process and glenoid fossa two-dimensional segmentation model to obtain a condyle process area and glenoid fossa area two-dimensional segmentation result; calculating condyle process form information, glenoid fossa form information and position information of the condyle process in the glenoid fossa based on the two-dimensional segmentation result of the condyle process area and the glenoid fossa area; wherein the mandible three-dimensional segmentation model and the condylar process and glenoid fossa two-dimensional segmentation model are obtained by training a deep learning network. According to the invention, automatic and accurate measurement of the three-dimensional shape and position of the condyle process and the glenoid fossa of the temporal-mandibular joint can be realized.

Description

Technical Field

[0001] The present application relates to the field of image processing technology, and in particular to a method, device and medium for measuring the three-dimensional morphology of a temporomandibular joint. Background Art

[0002] The temporomandibular joint (TMJ) is composed of the mandibular condyle, the articular surface of the temporal bone, the articular disc, the joint capsule and the joint ligaments. It is one of the most complex joints in the body. Temporomandibular joint diseases are common and frequently occurring diseases in the oral and maxillofacial region. Changes in the morphology and position of the temporomandibular joint may lead to an increase in the prevalence of temporomandibular joint diseases.

[0003] At present, quantitative measurement based on cone beam computed tomography (CBCT) images has become the most reliable method for evaluating the morphology and position of the condyle and fossa of the temporomandibular joint. As a three-dimensional imaging technology, CBCT can effectively avoid the problem of image overlap. It has the advantages of high spatial resolution, low radiation dose, and the ability to reconstruct images at any angle and in multiple planes. However, the three-dimensional quantitative evaluation of the condyle and fossa based on CBCT images mostly uses manual measurement methods, and there are still the following problems: 1) Most physicians lack or have not received systematic standardized training in oral and maxillofacial medical imaging. In addition, the relative scarcity of physicians for temporomandibular joint disease increases the difficulty of temporomandibular joint evaluation, and the accuracy of measurement is greatly affected by the physician's experience; 2) The measurement methods and measurement indicators are not unified, and there is no measurement tool with accurate measurement and good repeatability; 3) When the condyle is not fully developed or there are pathological conditions such as bone wear, bone hyperplasia and sclerosis, the quantitative evaluation of the condyle and fossa becomes more difficult; 4) The data volume is small, the measurement is cumbersome and time-consuming, and the consistency and repeatability are relatively poor. Summary of the invention

[0004] The purpose of this application is to provide a method, device and medium for measuring the three-dimensional morphology of the temporomandibular joint, which can realize automatic and accurate measurement of the three-dimensional morphology and position of the condyle and fossa of the temporomandibular joint.

[0005] To achieve the above objectives, this application provides the following solutions: In a first aspect, the present application provides a method for measuring the three-dimensional morphology of a temporomandibular joint, comprising: Acquire CBCT data of target temporomandibular joint; Inputting the target temporomandibular joint CBCT data into a mandibular three-dimensional segmentation model to obtain a mandibular three-dimensional segmentation result; Based on the mandibular three-dimensional segmentation result and the target temporomandibular joint CBCT data, determining a central position area image of the temporomandibular joint; Inputting the image of the central position area of ​​the temporomandibular joint into the condyle and fossa two-dimensional segmentation model to obtain the two-dimensional segmentation results of the condyle area and fossa area; Based on the two-dimensional segmentation results of the condyle region and the fossa region, calculating the three-dimensional morphological information of the temporomandibular joint; the three-dimensional morphological information of the temporomandibular joint includes the morphological information of the condyle, the morphological information of the fossa, and the position information of the condyle in the fossa; Among them, the three-dimensional segmentation model of the mandible is obtained by training the first preset deep learning network using the temporomandibular joint CBCT sample data set; the two-dimensional segmentation model of the condyle and fossa is obtained by training the second preset deep learning network using the temporomandibular joint center position area sample data set.

[0006] In a second aspect, the present application provides a computer device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement a method for measuring the three-dimensional morphology of the temporomandibular joint.

[0007] In a third aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements a method for measuring the three-dimensional morphology of the temporomandibular joint.

[0008] According to the specific embodiments provided by the present application, the present application has the following technical effects: the present application provides a method, device and medium for measuring the three-dimensional morphology of the temporomandibular joint, inputs the target temporomandibular joint CBCT data into the three-dimensional segmentation model of the mandible to obtain the three-dimensional segmentation result of the mandible, obtains the image of the central position area of ​​the temporomandibular joint from the segmentation result and inputs it into the two-dimensional segmentation model of the condyle and fossa to obtain the two-dimensional segmentation results of the condyle area and the fossa area, and then calculates the condyle morphology information, fossa morphology information and the position information of the condyle in the fossa according to the two-dimensional segmentation results of the condyle area and the fossa area. The above-mentioned whole process does not require the manual judgment of the doctor, is automatically completed, will not be affected by the doctor's experience, and the process is a unified process with good repeatability. In the above process, the three-dimensional segmentation model of the mandible and the two-dimensional segmentation model of the condyle and fossa were used to process the target temporomandibular joint CBCT data. These two models are deep learning networks based on artificial intelligence technology and have been trained in advance to quickly output accurate results. Therefore, the present application can quickly, automatically and accurately measure the condyle morphology information, fossa morphology information and the position information of the condyle in the fossa, that is, the three-dimensional morphology and position information of the condyle and fossa of the temporomandibular joint. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0010] Figure 1 This is a diagram of the application environment of the method for measuring the three-dimensional shape of the temporomandibular joint in one embodiment of the present application.

[0011] Figure 2 Schematic diagram of the process of measuring the three-dimensional morphology of the temporomandibular joint in one embodiment of the present application.

[0012] Figure 3 Schematic diagram of the measurement of condylar length and condylar width.

[0013] Figure 4 Schematic diagram of the measurement of condylar head height.

[0014] Figure 5 Schematic diagram of the condylar height measurement.

[0015] Figure 6 Schematic diagram of the measurement of glenoid fossa width, glenoid fossa depth and articular tubercle inclination.

[0016] Figure 7 Schematic diagram of the measurement of the anterior joint space, supra-articular space and posterior joint space.

[0017] Figure 8 Schematic diagram of the measurement of intra-articular space, mid-articular space and extra-articular space.

[0018] Fig. 9 , Fig.10 and Fig.11 They are respectively an original image in the temporomandibular joint CBCT sample data set to be used, the corresponding segmented image, and a schematic diagram of the 3D segmented image.

[0019] Fig.12 and Fig.13 They are respectively a coronal original image and a corresponding coronal two-dimensional segmentation image in the sample data set of the central position area of ​​the temporomandibular joint to be used.

[0020] Fig.14 and Fig.15 They are respectively a sagittal original image and a corresponding sagittal two-dimensional segmentation image in the sample data set of the central position area of ​​the temporomandibular joint to be used.

[0021] Fig.16 A schematic diagram of the structure of a computer device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0022] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0023] This application combines artificial intelligence technology to achieve automatic and precise measurement of the morphology and position of the condyle and fossa of the temporomandibular joint, and can serve as a powerful auxiliary tool to help dentists diagnose diseases, formulate treatment plans and evaluate treatment effects.

[0024] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0025] The method for measuring the three-dimensional morphology of the temporomandibular joint provided in the embodiment of the present application can be applied to Figure 1 In the application environment shown. Among them, the terminal 102 communicates with the server 104 through the network. The data storage system can store the data that the server 104 needs to process. The data storage system can be set up separately, integrated on the server 104, or placed on the cloud or other servers. The terminal 102 can send the target temporomandibular joint CBCT data to the server 104. After receiving it, the server 104 inputs it into the mandibular three-dimensional segmentation model to obtain the mandibular three-dimensional segmentation result, and determines the temporomandibular joint center position area image based on the mandibular three-dimensional segmentation result and the target temporomandibular joint CBCT data. The temporomandibular joint center position area image is input into the condyle and fossa two-dimensional segmentation model to obtain the condyle area and fossa area two-dimensional segmentation results, and finally, based on the condyle area and fossa area two-dimensional segmentation results, the condyle morphology information, fossa morphology information and the position information of the condyle in the fossa are calculated. In addition, the mandibular three-dimensional segmentation model and the condyle and fossa two-dimensional segmentation model are trained and saved in the server. The server 104 can feed back the obtained condylar morphology information, glenoid fossa morphology information and condylar position information in the glenoid fossa to the terminal 102. In addition, in some embodiments, the method for measuring the three-dimensional morphology of the temporomandibular joint can also be implemented by the server 104 or the terminal 102 alone.

[0026] The terminal 102 may be, but is not limited to, various desktop computers, laptop computers, smart phones, tablet computers, and IoT devices. The server 104 may be implemented as an independent server or a server cluster consisting of multiple servers, or may be a cloud server.

[0027] In an exemplary embodiment, Figure 2 As shown, a method for measuring the three-dimensional morphology of the temporomandibular joint is provided. The method is executed by a computer device, and can be executed by a computer device such as a terminal or a server alone, or by a terminal and a server together. In the embodiment of the present application, the method is applied to Figure 1 The server 104 in the example is used for explanation, and the steps include the following steps 201 to 205.

[0028] Step 201, acquiring target temporomandibular joint CBCT data.

[0029] Step 202, input the target temporomandibular joint CBCT data into the mandibular 3D segmentation model to obtain the mandibular 3D segmentation result; wherein the mandibular 3D segmentation model is obtained by training a first preset deep learning network using the temporomandibular joint CBCT sample data set. Through the processing of the above mandibular 3D segmentation model, the center position of the temporomandibular joint can be quickly found, and the influence of other redundant skull structures can be removed.

[0030] Step 203, based on the mandibular 3D segmentation result and the target temporomandibular joint CBCT data, determine the temporomandibular joint center position area image; specifically, based on the mandibular 3D segmentation result, all the corresponding temporomandibular joint center position area images are cut out from the target temporomandibular joint CBCT data, that is, the temporomandibular joint center position area image. The temporomandibular joint center position area image specifically includes three slice images, namely, the image of the maximum axial position of the condyle, the image of the corrected condyle center oblique sagittal position, and the image of the corrected condyle center oblique coronal position.

[0031] Step 204, input the image of the central position area of ​​the temporomandibular joint into the condyle and fossa two-dimensional segmentation model to obtain the two-dimensional segmentation results of the condyle area and fossa area; wherein the condyle and fossa two-dimensional segmentation model is obtained by training the second preset deep learning network using the sample data set of the central position area of ​​the temporomandibular joint. Through the processing of the above condyle and fossa two-dimensional segmentation model, the mask of the condyle and fossa on each slice image of the central position area image of the temporomandibular joint can be obtained, and the condyle area and fossa area are displayed on the corresponding mask.

[0032] Step 205, based on the two-dimensional segmentation results of the condyle area and the fossa area, calculate the three-dimensional morphological information of the temporomandibular joint; the three-dimensional morphological information of the temporomandibular joint includes condyle morphological information, fossa morphological information and position information of the condyle in the fossa; wherein, the condyle morphological information includes condyle length, condyle width, condylar head height and condyle height; the fossa morphological information includes fossa width, fossa depth and articular tubercle inclination; the position information of the condyle in the fossa includes anterior articular space, supra-articular space, posterior articular space, intra-articular space, mid-articular space and extra-articular space.

[0033] In a specific application example, in step 205, based on the two-dimensional segmentation results of the condyle region and the glenoid fossa region, the condyle morphology information is calculated, including the following steps (11)-(14).

[0034] (11) Corresponding to the maximum axis of the condyle, the innermost point of the condyle and the outermost point of the condyle are connected to obtain the condyle length.

[0035] (12) Corresponding to the maximum axis of the condyle, a line passing through the midpoint of the line connecting the innermost point of the condyle and the outermost point of the condyle and perpendicular to the inner and outer diameters of the condyle is obtained to obtain the anterior point of the condyle and the posterior point of the condyle, and the anterior point of the condyle and the posterior point of the condyle are connected to obtain the condyle width.

[0036] (13) Corresponding to the oblique sagittal position of the center of the corrected condyle, a tangent is drawn through the lowest point of the mandibular sigmoid notch, and a vertical line is drawn from the apex of the condyle to the tangent to obtain the condyle height.

[0037] (14) Corresponding to the corrected condylar center oblique coronal position, the innermost point of the condyle and the outermost point of the condyle are connected to obtain an inner-outer line, and a vertical line is drawn from the condylar apex to the inner-outer line to obtain the condylar head height.

[0038] In a specific application example, in step 205, based on the two-dimensional segmentation results of the condyle region and the glenoid fossa region, the glenoid fossa morphology information is calculated, including the following steps (21)-(23).

[0039] (21) Connect the lowest point of the articular tubercle with the lowest point of the posterior wall of the glenoid fossa to obtain the glenoid fossa width.

[0040] (22) Draw a perpendicular line through the apex of the glenoid fossa to the line connecting the lowest point of the articular tubercle and the lowest point of the posterior wall of the glenoid fossa to obtain the depth of the glenoid fossa.

[0041] (23) A first connecting line is made through the apex of the articular fossa and the lowest point of the articular tubercle, and the angle between the first connecting line and the orbitoauricular plane is taken as the inclination of the articular tubercle.

[0042] In a specific application example, in step 205, based on the two-dimensional segmentation results of the condyle region and the fossa region, the position information of the condyle in the fossa is calculated, including the following steps (31) to (32).

[0043] (31) Corresponding to the oblique sagittal position of the center of the corrected condyle, the anterior point of the condyle is connected with the anterior point of the fossa to obtain the anterior joint space; the apex of the condyle is connected with the apex of the fossa to obtain the supra-articular space; the posterior point of the condyle is connected with the posterior point of the fossa to obtain the posterior joint space.

[0044] (32) Corresponding to the corrected condylar center oblique coronal position, connect the point inside the articular fossa and the inner 1 / 4 point of the condylar apex to obtain the intra-articular space; connect the middle point of the articular fossa and the middle point of the condylar apex to obtain the mid-articular space; connect the point outside the articular fossa and the outer 1 / 4 point of the condylar apex to obtain the extra-articular space.

[0045] The above content is the measurement process of the 3D morphology of the temporomandibular joint. In practical applications, before measurement, it is necessary to construct and prepare the 3D segmentation model of the mandible and the 2D segmentation model of the condyle and the fossa. In the process of construction and preparation, it is first necessary to prepare the temporomandibular joint CBCT sample data set and the temporomandibular joint center position area sample data set, the first preset deep learning network and the second preset deep learning network.

[0046] Among them, any temporomandibular joint CBCT sample data in the temporomandibular joint CBCT sample data set includes historical temporomandibular joint CBCT data and corresponding mandibular three-dimensional segmentation result label. Any temporomandibular joint center position area sample data in the temporomandibular joint center position area sample data set includes historical temporomandibular joint center position area image and corresponding condyle segmentation label and fossa region segmentation label.

[0047] The acquisition process of historical temporomandibular joint CBCT data includes: when the patient takes a temporomandibular joint CBCT examination, the orbital-auricular plane is kept parallel to the ground plane. After the CBCT image scan is completed, the original DICOM data is imported into SmartVPro software. Each patient's temporomandibular joint CBCT can be collected as a historical data, and a total of 150 cases or more can be collected. Relevant technicians can adjust as needed. After obtaining multiple historical data, they are screened according to the following inclusion and exclusion criteria to obtain the final multiple historical temporomandibular joint CBCT data.

[0048] Inclusion criteria: ① clear images without motion artifacts or sclerosis artifacts; ② normal bone structure of condyle and glenoid fossa.

[0049] Exclusion criteria: ① incomplete or unclear images of the condyle and fossa; ② a history of temporomandibular joint tumors, trauma, ankylosis, or systemic diseases; ③ a history of temporomandibular joint trauma and surgery; ④ systemic diseases involving the condyle, such as rheumatoid arthritis, systemic lupus erythematosus, etc.

[0050] On this basis, for each historical temporomandibular joint CBCT data, the medial-lateral long axis of the condyle is determined at the level of the maximum cross section of the condyle in the axial position, and this level is the maximum axial position of the condyle; the sagittal position is adjusted to be perpendicular to the long axis of the condyle, and the central level of the sagittal position is the corrected condyle center oblique sagittal position; the coronal position is adjusted to be parallel to the long axis of the condyle, and the central level of the coronal position is the corrected condyle center oblique coronal position. The slice images of the above three slice positions constitute the historical temporomandibular joint center position regional image, that is, the historical temporomandibular joint center position regional image includes the images of the maximum axial position of the condyle, the corrected condyle center oblique sagittal position, and the corrected condyle center oblique coronal position.

[0051] The mandibular 3D segmentation result label is obtained by segmenting the mandibular 3D structure using mimics research 19.0. The segmentation result based on the segmentation includes the mandible and teeth. The 3D medical image processing software Labelme is used for 2D segmentation of the condyle and fossa regions. Since the accurate assessment of the 3D structure and morphology of the temporomandibular joint depends on the accurate segmentation of the mandible, condyle and fossa regions, the physicians are uniformly trained on the software usage and the 3D segmentation of the mandible and the 2D segmentation of the condyle and fossa before segmentation to ensure the accuracy of the segmentation.

[0052] After the two-dimensional segmentation results of the condyle and glenoid fossa are known, Figure 3 As shown, the condylar length and condylar width were measured, where the condylar length was the distance between the innermost point and the outermost point of the condyle at the maximum axial plane of the condyle, corresponding to Figure 3 AB in the figure; the condylar width is the distance between the anterior and posterior points of the condyle, at the maximum axial plane of the condyle, through the midpoint of the line connecting the innermost point of the condyle with the outermost point of the condyle and perpendicular to the inner and outer diameters of the condyle, corresponding to Figure 3 CD in.

[0053] Can be as Figure 4 As shown, the condylar head height is measured, where the condylar head height is: the vertical distance from the condylar vertex to the innermost point and the outermost point of the condyle on the oblique coronal image of the corrected condyle center, corresponding to Figure 4 HI in.

[0054] Can be as Figure 5As shown, the condylar height is measured, where the condylar height is: the vertical distance from the condylar vertex to the tangent line at the lowest point of the mandibular sigmoid notch on the oblique sagittal image of the corrected condylar center, corresponding to Figure 5 JK in.

[0055] Can be as Figure 6 As shown, the glenoid fossa width, glenoid fossa depth and articular tubercle inclination are measured, wherein the glenoid fossa width is the distance between the lowest point of the articular tubercle and the lowest point of the posterior wall of the glenoid fossa, corresponding to Figure 6 NM in the glenoid cavity; the depth of the glenoid cavity is the shortest distance between the glenoid cavity vertex perpendicular to the lowest point of the articular tubercle and the lowest point of the posterior wall of the glenoid cavity, corresponding to Figure 6 The SO in the articular tubercle is the angle between the first connecting line between the apex of the articular fossa and the lowest point of the articular tubercle and the orbitoauricular plane, corresponding to Figure 6 The angle ∠α in .

[0056] Can be as Figure 7 As shown, the anterior joint space, supra-articular space and posterior joint space are measured, wherein the anterior joint space corresponds to Figure 7 Q1-Q2, denoted as Q; the gap above the joint corresponds to Figure 7 S1-S2, denoted as S; the posterior joint space corresponds to Figure 7 In a specific application, the position of the condyle can be evaluated according to the method of Pullinger and Hollender: linear percentage = (PQ) / (P + Q) × 100%, such as linear percentage <-12%, indicating that the condyle is posterior; linear percentage > +12%, indicating that the condyle is anterior; linear percentage between -12% and +12%, indicating that the condyle is centered.

[0057] Can be as Figure 8 As shown, the intra-articular space, the middle joint space and the extra-articular space are measured, wherein the intra-articular space corresponds to Figure 8 V1-V2 in the joint is denoted as V; the gap in the joint corresponds to Figure 8 T1-T2 in the joint is denoted as T; the extra-articular space corresponds to Figure 8 U1-U2 in is denoted as U.

[0058] In addition, based on Figure 7 and Figure 8 The process of determining the condyle apex, the glenoid fossa apex, etc. can be known, specifically: Figure 7In the middle, draw a horizontal line Line1 parallel to the orbitoauricular plane and tangent to the upper edge of the articular fossa at the apex S2 of the articular fossa. Draw a tangent to the anterior edge of the condyle through the apex S2 of the articular fossa and tangent to the anterior point Q1 of the condyle. Draw a tangent to the posterior edge of the condyle and tangent to the posterior point P1 of the condyle. A perpendicular line Line2 through the apex S2 of the articular fossa intersects with the upper edge of the condyle at the apex S1 of the condyle. Draw a perpendicular line to the tangent to the anterior edge of the condyle through the anterior point Q1 of the condyle and intersects with the anterior edge of the articular fossa at the anterior point Q2. Draw a perpendicular line to the tangent to the posterior edge of the condyle through the posterior point P1 of the condyle and intersects with the posterior edge of the articular fossa at the posterior point P2 of the articular fossa. Figure 8 In the figure, the line connecting the innermost point G of the condyle and the outermost point F of the condyle is recorded as Line3, and a perpendicular line is drawn through the midpoint R of Line3, recorded as Line4. Line4 intersects with the superior edge of the condyle at the middle point T1 of the condyle apex, and intersects with the articular fossa at the middle point T2 of the articular fossa. The medial angle bisector of Line3 and Line4 intersects with the inner edge of the condyle at the inner 1 / 4 point V1 of the condyle apex, and intersects with the articular fossa at the inner point V2 of the articular fossa. The lateral angle bisector of Line3 and Line4 intersects with the outer edge of the condyle at the outer 1 / 4 point U1 of the condyle apex, and intersects with the articular fossa at the outer point U2 of the articular fossa.

[0059] In a specific application, the first preset deep learning network is a UNet model. Correspondingly, the training process of the mandibular three-dimensional segmentation model includes the following steps (41)-(42).

[0060] (41) Performing data expansion processing and data enhancement processing on the temporomandibular joint CBCT sample dataset to obtain a stand-by temporomandibular joint CBCT sample dataset; wherein the expansion method can adopt random cropping, random rotation, horizontal flipping, vertical flipping and contrast adjustment, and the data enhancement method can adopt an online enhancement mode to improve the generalization of the model.

[0061] (42) The standby temporomandibular joint CBCT sample data set is input into the UNet model for 3D segmentation, and the mandibular three-dimensional segmentation model is obtained through iterative optimization training. Among them, a lightweight mandibular segmentation model detection model, namely the UNet model, is built based on the Pytorch deep learning framework; secondly, the standby temporomandibular joint CBCT sample data set is divided into a training set, a validation set and a test set according to a ratio of 3:1:1, and then the training set is input into the UNet3D positioning model to output the predicted position heat map, so as to perform model training to achieve rough positioning and find the location of the temporomandibular joint. The above training process uses a focal loss function to ensure sensitivity to position, which is defined as: .

[0062] Among them, Loss is the value of the loss function, is the number of samples in the temporomandibular joint CBCT sample dataset to be used, To predict a pixel, is 0.5, To mark the pixels, is 0.5, x, y, z are coordinate values.

[0063] In a practical application, the image sizes of the training set, validation set, and test set can be unified to 72×72×72. Fig. 9 , Fig.10 and Fig.11 As shown in FIG. 1 , they are an original image, a corresponding segmented image, and a 3D segmented image in the temporomandibular joint CBCT sample data set to be used. Fig.11 After viewing the image shown, the maximum axial plane of the condyle can be located and the maximum axial-coronal plane can be output.

[0064] In another specific application, the second preset deep learning network is a nnUNet model, and correspondingly, the training process of the condyle and glenoid fossa two-dimensional segmentation model includes the following steps (51)-(52).

[0065] (51) Performing data enhancement processing on the sample data set of the central position area of ​​the temporomandibular joint to obtain a sample data set of the central position area of ​​the temporomandibular joint to be used. In practical applications, the process of obtaining the sample data set of the central position area of ​​the temporomandibular joint is as follows: after the condyle and fossa are roughly positioned using the coarse positioning model of the condyle and fossa, this process corresponds to steps (41) to (42) in the above text, the image of the corresponding central area of ​​the temporomandibular joint is cropped out, and then the condyle and fossa are annotated as a data set of two-dimensional segmentation of the condyle and fossa, thereby obtaining a sample data set of the central position area of ​​the temporomandibular joint. In addition, the data enhancement processing in this step includes random rotation, random scaling, adding noise, etc.

[0066] (52) Input the sample data set of the central position area of ​​the temporomandibular joint to be used into the nnUNet model for 2D segmentation training to obtain a two-dimensional segmentation model of the condyle and fossa. In an application example, the set is divided into training set, validation set and test set in a ratio of 3:1:1, and the nnUNet model is selected to train the 2D segmentation model. The training process uses cross entropy as the loss function to achieve two-dimensional segmentation of the condyle and fossa on 2D slices, obtain the mask of the condyle and fossa on each slice, and display the two-dimensional segmentation results of the condyle area and fossa area on the corresponding mask. Fig.12 and Fig.13 As shown in FIG. 1 , there is a coronal original image and a corresponding coronal two-dimensional segmented image in the sample data set of the central position area of ​​the temporomandibular joint to be used, such as Fig.14 and Fig.15As shown, there is a sagittal original image and a corresponding sagittal two-dimensional segmentation image in the sample data set of the central position area of ​​the temporomandibular joint to be used.

[0067] After the training of the above-mentioned condyle and glenoid fossa two-dimensional segmentation model is completed, the segmentation results can be compared with the gold standard to calculate the accuracy of the segmentation and the accuracy of the measurement indicators. If the requirements are not met, it is necessary to reselect the temporomandibular joint CBCT sample data set and the temporomandibular joint center position area sample data set for training.

[0068] In summary, this application aims to establish a CBCT image automatic measurement system for the three-dimensional morphology and position of the condyle and fossa of the temporomandibular joint based on deep learning artificial intelligence technology, so as to achieve three-dimensional automatic and accurate measurement of the temporomandibular joint. The establishment of an automatic measurement model for the three-dimensional morphology and position of the condyle and fossa of the temporomandibular joint has significant beneficial effects in improving measurement accuracy, achieving rapid measurement, supporting personalized treatment, auxiliary diagnosis and scientific research, and promoting interdisciplinary cooperation. This will provide strong technical support and guarantee for the diagnosis and treatment of diseases related to the temporomandibular joint.

[0069] In an exemplary embodiment, a computer device is provided. The computer device may be a server or a terminal. The internal structure diagram thereof may be as follows: Fig.16 As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a method for measuring the three-dimensional morphology of the temporomandibular joint is implemented.

[0070] Those skilled in the art will understand that Fig.16The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components. In an exemplary embodiment, a computer device is provided, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps in the above-mentioned method embodiments.

[0071] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0072] In an exemplary embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0073] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0074] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0075] The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a distributed database based on blockchain, etc., but is not limited thereto. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but is not limited thereto.

[0076] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0077] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A method for measuring the three-dimensional morphology of a temporomandibular joint, characterized in that: Methods include: Acquire CBCT data of target temporomandibular joint; Inputting the target temporomandibular joint CBCT data into a mandibular three-dimensional segmentation model to obtain a mandibular three-dimensional segmentation result; Based on the mandibular three-dimensional segmentation result and the target temporomandibular joint CBCT data, determining a central position area image of the temporomandibular joint; Inputting the image of the central position area of ​​the temporomandibular joint into the condyle and fossa two-dimensional segmentation model to obtain the two-dimensional segmentation results of the condyle area and fossa area; Calculating the three-dimensional morphological information of the temporomandibular joint based on the two-dimensional segmentation results of the condyle region and the glenoid fossa region; The three-dimensional morphological information of the temporomandibular joint includes the morphological information of the condyle, the morphological information of the fossa, and the position information of the condyle in the fossa; Wherein, the mandibular three-dimensional segmentation model is obtained by training a first preset deep learning network using a temporomandibular joint CBCT sample data set; The condyle and glenoid fossa two-dimensional segmentation model is obtained by training a second preset deep learning network using a sample data set of the temporomandibular joint center area.

2. The method for measuring the three-dimensional morphology of the temporomandibular joint according to claim 1, characterized in that: The condyle morphology information includes condyle length, condyle width, condyle head height and condyle height; the fossa morphology information includes fossa width, fossa depth and articular tubercle inclination; the condyle position information in the fossa includes the anterior articular space, supra-articular space, posterior articular space, intra-articular space, mid-articular space and extra-articular space.

3. The method for measuring the three-dimensional morphology of the temporomandibular joint according to claim 1, characterized in that: Any temporomandibular joint CBCT sample data in the temporomandibular joint CBCT sample data set includes historical temporomandibular joint CBCT data and corresponding mandibular three-dimensional segmentation result labels; The first preset deep learning network is a UNet model; The training process of the mandibular 3D segmentation model includes: Performing data expansion processing and data enhancement processing on the temporomandibular joint CBCT sample data set to obtain a temporomandibular joint CBCT sample data set to be used; The temporomandibular joint CBCT sample data set to be used is input into the UNet model for 3D segmentation, and a mandibular three-dimensional segmentation model is obtained through iterative optimization training.

4. The method for measuring the three-dimensional morphology of the temporomandibular joint according to claim 1, characterized in that: Any temporomandibular joint center position area sample data in the temporomandibular joint center position area sample data set includes a historical temporomandibular joint center position area image and a corresponding condylar region segmentation label and fossa region segmentation label; The second preset deep learning network is a nnUNet model; The training process of the condyle and glenoid fossa two-dimensional segmentation model includes: Performing data enhancement processing on the temporomandibular joint center position area sample data set to obtain a stand-by temporomandibular joint center position area sample data set; The sample data set of the central position area of ​​the temporomandibular joint to be used is input into the nnUNet model for 2D segmentation training to obtain a two-dimensional segmentation model of the condyle and the fossa.

5. The method for measuring the three-dimensional morphology of the temporomandibular joint according to claim 4, characterized in that: The historical images of the central position area of ​​the temporomandibular joint include images of the maximum axial position of the condyle, the oblique sagittal position of the corrected condyle center, and the oblique coronal position of the corrected condyle center; Based on the two-dimensional segmentation results of the condyle region and the glenoid fossa region, the condyle morphology information is calculated, including: Corresponding to the maximum axis of the condyle, connecting the innermost point of the condyle and the outermost point of the condyle to obtain the condyle length; Corresponding to the maximum axis of the condyle, a line passing through the middle point of the line connecting the innermost point of the condyle and the outermost point of the condyle and perpendicular to the inner and outer diameters of the condyle is obtained to obtain the anterior point of the condyle and the posterior point of the condyle, and the anterior point of the condyle is connected with the posterior point of the condyle to obtain the condyle width; Corresponding to the oblique sagittal position of the center of the corrected condyle, a tangent is made through the lowest point of the mandibular sigmoid notch, and a vertical line is made from the apex of the condyle to the tangent to obtain the height of the condyle; Corresponding to the corrected condylar center oblique coronal position, the innermost point of the condyle and the outermost point of the condyle are connected to obtain an inner-outer line, and a vertical line is drawn from the condylar apex to the inner-outer line to obtain the condylar head height.

6. The method for measuring the three-dimensional morphology of the temporomandibular joint according to claim 5, characterized in that: Based on the two-dimensional segmentation results of the condyle region and the glenoid fossa region, calculating the glenoid fossa morphology information includes: The lowest point of the articular tubercle was connected with the lowest point of the posterior wall of the glenoid fossa to obtain the glenoid fossa width; A vertical line is drawn through the apex of the glenoid fossa to the connecting line between the lowest point of the articular tubercle and the lowest point of the posterior wall of the glenoid fossa to obtain the depth of the glenoid fossa; A first connecting line is made through the apex of the articular fossa and the lowest point of the articular tubercle, and the angle between the first connecting line and the orbitoauricular plane is taken as the inclination of the articular tubercle.

7. The method for measuring the three-dimensional morphology of the temporomandibular joint according to claim 5, characterized in that: Based on the two-dimensional segmentation results of the condyle region and the fossa region, calculating the position information of the condyle in the fossa, including: Corresponding to the corrected condylar center oblique sagittal position, the anterior point of the condyle is connected with the anterior point of the fossa to obtain the anterior joint space; the apex of the condyle is connected with the apex of the fossa to obtain the supra-articular space; the posterior point of the condyle is connected with the posterior point of the fossa to obtain the posterior joint space; Corresponding to the corrected condylar center oblique coronal position, connect the inner point of the articular fossa and the inner 1 / 4 point of the condylar top to obtain the intra-articular space; connect the middle point of the articular fossa and the middle point of the condylar top to obtain the mid-articular space; connect the outer point of the articular fossa and the outer 1 / 4 point of the condylar top to obtain the extra-articular space.

8. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method for measuring the three-dimensional morphology of the temporomandibular joint as described in any one of claims 1-7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for measuring the three-dimensional morphology of the temporomandibular joint described in any one of claims 1 to 7 is implemented.

Images