Method for Deriving Head Measurement Parameters for Machine Learning-Based Orthodontic Diagnosis from 3D CBCT Images Taken in the Natural Head Position

Abstract

The present invention relates to a machine learning-based cephalometric parameter derivation method for orthodontic diagnosis, more particularly, to a machine learning-based cephalometric parameter derivation method for orthodontic diagnosis, which is capable of acquiring a subject's cephalometric image from cone beam computed tomography (CBCT) image data taken in a natural head position by applying a machine learning algorithm, and detecting a plurality of measurement points on the cephalometric image precisely and quickly to derive 13 diagnostic parameters for precise orthodontic diagnosis.

Description

Technical Field

[0001] The present invention relates to a method for deriving head measurement parameters for machine learning-based orthodontic diagnosis. More specifically, the present invention applies a machine learning algorithm and video analysis processing technology to obtain a head measurement image of a subject from video data obtained by cone beam computed tomography (CBCT) in a natural head position state, and in order to derive 13 parameters for precise orthodontic diagnosis, the present invention relates to a method for deriving head measurement parameters for orthodontic diagnosis based on machine learning that can accurately and quickly detect a plurality of measurement points on the head measurement image.

Background Art

[0002] Generally, a state in which the tooth alignment is disordered and the occlusion of the upper and lower teeth is incorrect is called malocclusion, and orthodontic treatment may be performed to correct such malocclusion to normal occlusion. On the other hand, in the precise diagnosis for orthodontics and the establishment of a treatment plan, an operation of detecting anatomically predetermined anatomical landmarks on the head measurement image of the subject is required.

[0003] Problems such as screen distortion (screen distortion) or blurriness in existing X-ray images are complemented, and recently, a three-dimensional CBCT image has been obtained from the head medical video data of a subject taken using a cone beam computer tomography (CBCT) device, and research has been conducted to detect measurement points for deriving parameters for orthodontic diagnosis based on this. In particular, in recent research, as a CBCT image analysis method, a Nasion True Vertical Plane (NTVP) perpendicular to the ground while passing through the lowest nasion point in the boundary region between the forehead and the nose, and a True Horizontal Plane (THP) horizontal to the ground while passing through the nasion point are used to grasp the anterior-posterior skeletal relationship and the degree of tooth protrusion.

[0004] Conventionally, in order to derive orthodontic diagnosis parameters, skilled operators such as medical personnel had to manually detect more than 50 measurement points on multiple CBCT images. However, in such a method, since the measurement point detection method and detection accuracy vary depending on the skill level of the operator, accurate orthodontic diagnosis is difficult, the measurement point detection time takes as long as about 30 minutes or more, and there is a problem of reduced medical treatment efficiency.

[0005] In order to solve such problems, by introducing a machine learning algorithm into the field of orthodontic diagnosis, a subject head measurement image is obtained from video data obtained by cone beam computed tomography (CBCT) in a state of natural head position, and a reduced number of parameters compared to the existing ones are derived for precise orthodontic diagnosis. At present, there is an urgent need for a method for deriving head measurement parameters for machine learning-based orthodontic diagnosis that can accurately and quickly detect multiple measurement points on the head measurement image to improve medical treatment efficiency.

Summary of the Invention

Problems to be Solved by the Invention

[0006] An object of the present invention is to provide a method for deriving head measurement parameters for machine learning-based orthodontic diagnosis, which applies a machine learning algorithm to obtain a subject head measurement image from video data obtained by cone beam computed tomography (CBCT) in a natural head position state, and accurately and quickly derives a plurality of measurement points on the head measurement image to derive 13 reduced parameters compared to the existing ones for precise orthodontic diagnosis and improve medical treatment efficiency.

Means for Solving the Problems

[0007] In order to solve the above problems, the present invention provides a method for deriving head measurement parameters for machine learning-based orthodontic diagnosis using head measurement images for diagnosis, which are extracted from video data obtained by 3D cone beam computed tomography (CBCT) of a subject's head in a natural head position state, and include a thalamic head image, a coronal head image, an oral panoramic image, and an anterior tooth cross-sectional image for the subject. The method includes detecting a plurality of measurement points for deriving 13 parameters for orthodontic diagnosis on the head measurement image based on a machine learning algorithm, and deriving 13 parameters corresponding to distances or angles between the plurality of detected measurement points. See The 13 parameters include parameters derived using the Nasion True Vertical Plane (NTVP), which is a vertical plane passing through the nasion, which is the lowest part in the boundary region between the forehead and the nose among the plurality of measurement points, or the True Horizontal Plane (THP), which is a horizontal plane passing through the nasion It is possible to provide a method for deriving head measurement parameters for orthodontic diagnosis, characterized by the above.

[0008] Here, the 13 parameters may include the protrusion degree of the maxilla, the protrusion degree of the mandible, the protrusion degree of the chin tip, the degree of displacement of the center of the mandible, the degree of displacement of the center of the maxillary central incisor, the degree of displacement of the center of the mandibular central incisor, the vertical distance from the nasion (the lowest part in the boundary region between the forehead and the nose) to the lower end point of the right canine tooth through the true horizontal plane (THP), the vertical distance from the THP to the lower end point of the left canine tooth, the vertical distance from the THP to the right maxillary first molar, the vertical distance from the THP to the left maxillary first molar, the inclination degree of the maxillary central incisor, the inclination degree of the mandibular central incisor, and the vertical inclination degree of the mandible with respect to the THP, in order to provide information for orthodontic diagnosis.

[0009] Further, the machine learning algorithm can divide the region provided between the frontal head and the rhinion on the thalamic head image, the region from the rhinion to the upper teeth, the region from the upper teeth to the menton of the mandible, and the region from the menton of the mandible to the articulare of the temporomandibular joint, and detect a plurality of measurement points for deriving the 13 parameters.

[0010] Then, the machine learning algorithm can divide the region from the frontal lobe to the rhinion on the coronal head image and the mandibular region, and detect a plurality of measurement points for deriving the 13 parameters.

[0011] In addition, the machine learning algorithm includes a process of applying an R-CNN (Region based convolutional neural networks) machine learning model to the oral panoramic image to sense individual regions of the entire teeth, a process of detecting tooth measurement points indicating the positions of the teeth for each of the sensed individual regions of the entire teeth, a process of analyzing the positions of the detected tooth measurement points to classify the entire teeth into upper teeth and lower teeth, a process of sequentially numbering the right upper teeth, left upper teeth, right lower teeth, and left lower teeth according to the horizontal distance from the facial midline to the detected tooth measurement points, and a process of analyzing the numbered teeth and detecting a plurality of measurement points for deriving parameters in specific teeth including incisors, canines, and first molars.

[0012] Then, in the thalamic head image, the machine learning algorithm detects point A (A point, A), which is the deepest part on the line connecting the Nasion, which is the lowest part in the boundary region between the forehead and the nose, the Anterior nasal spine in the maxilla, and the Prosthion in the maxillary anterior tooth alveolus. The protrusion of the maxilla may be derived by measuring the distance between the Nasion true vertical plane (NTVP), which is a vertical plane passing through the Nasion, and point A.

[0013] In addition, in the thalamic head image, the machine learning algorithm detects point B (B point, B), which is the deepest part on the line connecting the Nasion, which is the lowest part in the boundary region between the forehead and the nose, the Infradentale of the mandibular anterior teeth, and the Pog. The protrusion of the mandible may be derived by measuring the distance between the Nasion true vertical plane (NTVP), which is a vertical plane passing through the Nasion, and point B.

[0014] In addition, the machine learning algorithm may detect, in the thalamic head image, the Nasion, which is the lowest part in the boundary region between the forehead and the nose, and the Pogonion (Pog), which is the most forward point of the mandible, and derive the protrusion degree of the chin tip by measuring the distance between the Nasion true vertical plane (NTVP), which is a vertical plane passing through the Nasion, and the Pogonion.

[0015] In addition, the machine learning algorithm may detect, in the coronal head image, the Nasion, which is the lowest part in the boundary region between the forehead and the nose, and the Menton, which is the lowest point of the mandible, and derive the degree of displacement of the center of the mandible by measuring the distance between the Nasion true vertical plane (NTVP), which is a vertical plane passing through the Nasion, and the Menton.

[0016] In addition, the machine learning algorithm may detect, in the coronal head image, the Nasion, which is the lowest part in the boundary region between the forehead and the nose, and the center point of the maxillary central incisor in the oral panoramic image, and derive the degree of displacement of the center of the maxillary central incisor by measuring the distance between the Nasion true vertical plane (NTVP), which is a vertical plane passing through the Nasion, and the center point of the maxillary central incisor.

[0017] In addition, the machine learning algorithm may detect, in the coronal head image, the Nasion, which is the lowest part in the boundary region between the forehead and the nose, and the center point of the mandibular central incisor on the oral panoramic image, and derive the degree of displacement of the center of the mandibular central incisor by measuring the distance between the Nasion true vertical plane (NTVP), which is a vertical plane passing through the Nasion, and the center point of the mandibular central incisor.

[0018] In addition, the machine learning algorithm may detect the Nasion, which is the lowest part in the boundary region between the forehead and the nose in the coronal head image, and the lower end point of the right canine tooth on the oral panoramic image, and derive the vertical distance between the True horizontal plane (THP), which is the horizontal plane passing through the Nasion, and the lower end point of the right canine tooth.

[0019] In addition, the machine learning algorithm may detect the Nasion, which is the lowest part in the boundary region between the forehead and the nose in the coronal head image, and the lower end point of the left canine tooth in the oral panoramic image, and derive the vertical distance between the True horizontal plane (THP), which is the horizontal plane passing through the Nasion, and the lower end point of the left canine tooth.

[0020] In addition, the machine learning algorithm may detect the Nasion, which is the lowest part in the boundary region between the forehead and the nose in the coronal head image, and the lower end point of the right first maxillary molar on the oral panoramic image, and derive the vertical distance from the THP to the right first maxillary molar based on the distance between the True horizontal plane (THP), which is the horizontal plane passing through the Nasion, and the lower end point of the right first maxillary molar.

[0021] In addition, the machine learning algorithm may detect the Nasion, which is the lowest part in the boundary region between the forehead and the nose in the coronal head image, and the lower end point of the left first maxillary molar on the oral panoramic image, and derive the vertical distance from the THP to the left first maxillary molar based on the distance between the True horizontal plane (THP), which is the horizontal line passing through the Nasion, and the lower end point of the left first maxillary molar.

[0022] Further, the machine learning algorithm detects the Nasion, which is the lowest part in the boundary region between the forehead and the nose in the coronal plane head image, and the upper and lower endpoints of the maxillary anterior teeth on the anterior tooth cross-sectional image, and the inclination of the maxillary central incisor may be derived from the angle of the vector connecting between the THP (True horizontal plane), which is the horizontal plane passing through the Nasion, and the upper and lower endpoints of the maxillary anterior teeth.

[0023] Further, the machine learning algorithm detects the Menton, which is the lowest point of the mandible in the thalamic plane head image, the Gonion (angle of the jaw; the point of maximum curvature of the mandible), and the upper and lower endpoints of the mandibular anterior teeth in the anterior tooth cross-sectional image, and the inclination of the mandibular central incisor may be derived from the angle between the MeGo line connecting the Menton and the Gonion and the vector connecting the upper and lower endpoints of the mandibular anterior teeth.

[0024] Further, the machine learning algorithm detects the Nasion, which is the lowest part in the boundary region between the forehead and the nose in the thalamic plane head image, the Menton, which is the lowest point of the mandible, and the Gonion, which is the point of maximum curvature of the mandible, and the vertical inclination of the mandible with respect to the THP may be derived from the angle between the THP (True horizontal plane) passing through the Nasion and the MeGo line connecting the Menton and the Gonion.

[0025] The present invention may further include a step of Analysis determining the facial shape or occlusion state of the subject corresponding to the 13 derived parameters.

[0026] Here, when determining the occlusion state of the subject corresponding to the 13 derived parameters, Analysis states where the anterior-posterior occlusion state between the maxilla and the mandible is in a relatively normal range, a state where the maxilla protrudes more than the mandible, and a state where the mandible protrudes more than the maxilla are classified respectively Analysis and can be achieved.

[0027] In addition, when shaping the subject's facial shape corresponding to the derived 13 parameters, Analysis classify the state where the length of the facial part is within the normal range, the state where the length of the facial part is shorter than the normal range, and the state where the length of the facial part is longer than the normal range, respectively. Analysis This can be done.

[0028] Note that the present invention uses, as input data, a head diagnostic head measurement image extracted at the stage of acquiring a diagnostic head measurement image of a method for deriving head measurement parameters for orthodontic diagnosis, detects a plurality of the above-described diagnostic points as output data, and is programmed such that the step of deriving 13 parameters corresponding to the distances or angles between the detected plurality of diagnostic points is automatically performed, and is provided on a computing device or a computable cloud server. A parameter derivation program for orthodontic diagnosis can be provided.

Advantages of the Invention

[0029] According to the method for deriving head measurement parameters for machine learning-based orthodontic diagnosis according to the present invention, by applying a machine learning algorithm, head images in the sagittal plane and coronal plane directions, anterior tooth cross-sectional images, and oral panoramic images for a subject are extracted from video data obtained by cone beam computed tomography (CBCT) in the natural head position, and measurement points determined in advance for extracting parameters from the images are automatically detected, so that orthodontic diagnosis work can be processed very quickly at a level within about several tens of seconds.

[0030] According to the method for deriving cephalometric parameters for machine learning-based orthodontic diagnosis according to the present invention, in order to smoothly grasp the anteroposterior skeletal relationship and the degree of tooth protrusion on the thalamus plane based on the measurement points detected on the cone beam computed tomography (CBCT) cephalometric image taken in the natural head position, 13 diagnostic parameters reduced compared to the conventional method are selected and derived. Thereby, the machine learning algorithm for detecting the measurement points is simplified, and the time required for orthodontic diagnosis can be shortened.

[0031] In addition, according to the method for deriving cephalometric parameters for machine learning-based orthodontic diagnosis according to the present invention, by extracting a diagnostic cephalometric image using a machine learning algorithm and detecting measurement points on the image, 13 parameters are derived to automatically Analysis not only the function of determining the facial shape or occlusion state of the subject, but also the application range can be further expanded, such as automatically designing a customized dental orthosis for the subject corresponding to the derived parameters.

Brief Description of the Drawings

[0032]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

Embodiments for Carrying Out the Invention

[0033] Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. However, the present invention is not limited to the embodiments described herein and may be embodied in other forms. Rather, the embodiments introduced here are provided so that the disclosed content can be thorough and complete, and so that the inventive concept can be fully conveyed to those skilled in the art. The same reference numerals throughout the specification denote the same components.

[0034] FIG. 1 is a flowchart showing a method for deriving head measurement parameters for machine learning-based orthodontic diagnosis according to the present invention.

[0035] As shown in FIG. 1, the method for deriving machine learning-based head measurement parameters according to the present invention includes the steps of: obtaining diagnostic head measurement images including head images in the sagittal plane and coronal plane directions, head images in the vertical direction, oral panoramic images, and anterior tooth cross-sectional images for the subject from CBCT video data obtained by photographing with a dental cone-beam computed tomograph (CBCT) with the subject's head positioned in a natural head position (S100); detecting a plurality of measurement points for deriving 13 parameters for orthodontic diagnosis on the diagnostic head measurement images based on a machine learning algorithm (S200); and deriving 13 parameters corresponding to distances or angles between the plurality of detected measurement points (S300).

[0036] In addition, the method for deriving machine learning-based head measurement parameters according to the present invention may further include the step of Analysis characterizing the facial shape or occlusion state of the subject corresponding to the 13 derived parameters (S400).

[0037] FIG. 2 is a diagram showing the process of obtaining diagnostic head measurement images for a subject in the method for deriving machine learning-based head measurement parameters according to the present invention.

[0038] As shown in FIG. 2, in the step (S100) of obtaining the diagnostic head measurement images, a plurality of diagnostic head measurement images including a head image 20 in the sagittal plane direction divided left and right for the subject, a head image 30 in the coronal plane direction divided front and back for the subject, an oral panoramic image 40, and an anterior tooth cross-sectional image 50 can be obtained respectively from the CBCT video data 10 obtained by photographing with a cone-beam computed tomograph.

[0039] In the step (S100) of obtaining the diagnostic head measurement image, three-dimensional head medical video data for the entire region of the subject can be obtained using a dental cone beam computed tomography (CBCT) device. The CBCT video data 10 can meet the medical digital imaging and communications in medicine (DICOM) standard by a machine learning algorithm, and the diagnostic head measurement image can be obtained from the CBCT video data by an image extraction function input into the machine learning algorithm or an image extraction function of a general DICOM viewer.

[0040] Here, the CBCT video data 10 may be extracted as a diagnostic head measurement image including a thalamic plane head image 20, a coronal plane head image 30, an oral panoramic image 40, and an anterior tooth cross-sectional image 50. Among them, the thalamic plane head image 20 and the coronal plane head image 30 can be classified and extracted into a thalamic plane bone mode image 20a and a coronal plane bone mode image 30a, respectively, which project and show the inside of the skull bone tissue, and can also be classified into a thalamic plane depth mode image 20b and a coronal plane depth mode image 30b, which are modes that can show the external form considering the depth or density of the subject's skull bone tissue, and the head measurement image can be extracted.

[0041] Thereafter, in the step (S200) of detecting the plurality of measurement points, a plurality of measurement points for deriving parameters for orthodontic diagnosis may be automatically detected on the diagnostic head measurement image using a machine learning algorithm. Conventionally, a skilled person such as a dentist manually specified the measurement points on the diagnostic head measurement image. However, such a method has a problem that the accuracy deviates depending on the skill level of the operator, and the measurement point detection time takes as long as about 30 minutes to 1 hour, resulting in a decrease in medical treatment efficiency.

[0042] On the one hand, in the present invention, a plurality of measurement points determined in advance in the diagnostic head measurement image can be automatically detected by using a face profile automatic analysis model and a machine learning algorithm including R-CNN (Region based Convolutional Neural Network). As a result, in the step (S300) of deriving the 13 parameters, 13 parameters selected to correspond to the distances or angles defined between a plurality of measurement points are derived, and the facial state or oral state of the subject is sensed by using the derived 13 parameters, and an orthodontic diagnosis for the subject can be performed.

[0043] Conventionally, in order to derive orthodontic diagnosis parameters, a large number of more than about 50 head measurement points had to be detected from the diagnostic head measurement image. However, in the present invention, in order to efficiently Analysis perform the skeletal relationship before and after the thalamus, occlusion state, tooth protrusion degree, etc. of the subject, by strictly selecting 13 parameters, the number of measurement points to be detected for deriving them is epoch-makingly reduced, and thereby, the machine learning algorithm for embodying it is simplified, and the required time for orthodontic diagnosis can be shortened.

[0044] Instead of using the Wits analysis method, Rickett analysis method, or McNamara analysis method, etc., which have been widely used in the existing head measurement analysis method for the purpose of orthodontic diagnosis, the present inventors utilize the Nasion True vertical plane (NTVP), which is a vertical plane passing through the Nasion (N point), which is generally the starting point of the nasal column, or the True horizontal plane (THP), which is a horizontal plane passing through the Nasion, in the video taken in the Natural head position, and while significantly reducing the number of measurement points compared to the existing ones, the relationship between the anterior and posterior jaws of the subject can be easily Analysis determined. By devising a method and applying it to a machine learning algorithm, the process of detecting a plurality of measurement points can be performed quickly and efficiently.

[0045] FIG. 3 is a diagram showing the positions of a plurality of measurement points for deriving 13 parameters from a thalamic plane head image 20, a coronal plane head image 30, an oral panoramic image 40, and an anterior tooth cross-sectional image 50.

[0046] As shown in FIG. 3, in the measurement point detection step (S200), measurement points determined in advance for deriving the 13 parameters may be automatically detected by a machine learning algorithm and video analysis processing technology from the thalamic plane head image 20, the coronal plane head image 30, and the oral panoramic image 40 obtained in the diagnostic head measurement image acquisition step (S100) based on the machine learning algorithm.

[0047] Here, the 13 parameters may be derived as the protrusion degree of the maxilla, the protrusion degree of the mandible, the protrusion degree of the jaw tip, the degree of displacement of the center of the mandible, the degree of displacement of the center of the maxillary central incisor, the degree of displacement of the center of the mandibular central incisor, the vertical distance from the right canine lower end point to the true horizontal plane (THP) passing through the nasion, which is the lowest part in the boundary region between the forehead and the nose, the vertical distance from the left canine lower end point to the THP, the vertical distance from the right maxillary first molar to the THP, the vertical distance from the left maxillary first molar to the THP, the inclination of the maxillary central incisor, the inclination of the mandibular central incisor, and the vertical inclination of the mandible with respect to the THP in order to provide information for orthodontic diagnosis for the subject.

[0048] For this purpose, the 13 parameters may be defined by measurement point keypoints corresponding to distances or angles between a plurality of measurement points detected from the diagnostic head measurement image as shown in Table 1 below.

[0049]

Table 1

[0050] Referring to FIG. 3 and Table 1, a plurality of measurement points detected from each diagnostic head measurement image, measurement point key points corresponding to distances or angles between the plurality of measurement points, and 13 parameters derived therefrom will be described.

[0051] In the thalamic head image 20, the machine learning algorithm detects the A point (A point, A), which is the lowest part on the line connecting the Nasion (N), the Anterior nasal spine in the maxilla, and the Prosthion. The prominence of the maxilla, which is one of the 13 parameters, is derived by measuring the distance in the X-axis direction in the thalamic head image 20 between the Nasion true vertical plane (NTVP), which is a vertical plane passing through the Nasion (N), and the A point (A), and the anteroposterior relationship between the maxilla and the mandible of the subject can be Analysis determined.

[0052] Also, in the thalamic head image 20, the machine learning algorithm detects the B point (B point, B), which is the deepest part on the line connecting the Nasion (N), the Infradentale of the mandibular anterior teeth, and the Pog. The prominence of the maxilla, which is one of the 13 parameters, is derived by measuring the distance in the X-axis direction from the thalamic head image 20 between the Nasion true vertical plane (NTVP), which is a vertical plane passing through the Nasion (N), and the B point (B), and the anteroposterior relationship between the maxilla and the mandible of the subject can be Analysis determined.

[0053] In addition, the machine learning algorithm detects the Nasion (N), which is the lowest part in the boundary region between the forehead and the nose, from the thalamic head image 20, and the Pogonion (Pog) that protrudes most forward from the mandible. The prominence of the chin tip, which is one of the 13 parameters, is derived by measuring the distance in the X-axis direction in the thalamic head image 20 between the Nasion true vertical plane (NTVP), which is a vertical plane passing through the Nasion (N), and the Pogonion (Pog). The anteroposterior relationship between the maxilla and the mandible of the subject can be Analysis determined.

[0054] In addition, the machine learning algorithm detects the Nasion (N), which is the lowest part in the boundary region between the forehead and the nose, and the Menton (Me), which is the lowest point on the mandible, from the coronal head image 30. The degree of displacement of the center of the mandible, which is one of the 13 parameters, is derived by measuring the distance in the Y-axis direction in the coronal head image 30 between the Nasion true vertical plane (NTVP), which is a vertical plane passing through the Nasion (N), and the Menton (Me). The left and right occlusal relationship between the maxilla and the mandible of the subject can be Analysis determined.

[0055] In addition, the machine learning algorithm detects the Nasion (N), which is the lowest part in the boundary region between the forehead and the nose extracted from the coronal head image 30, and the vertical line passing through the center of the upper central incisor (UDM) from the oral panoramic image 40. The degree of displacement of the center of the upper central incisor, which is one of the 13 parameters, is derived by measuring the distance in the Y-axis direction in the coronal head image between the Nasion true vertical plane (NTVP), which is a vertical plane passing through the Nasion (N), and the vertical line passing through the center of the upper central incisor (UDM). The left and right occlusal relationship between the maxilla and the mandible of the subject can be Analysis determined.

[0056] In addition, the machine learning algorithm detects the nasion point (N), which is the lowest part in the boundary region between the forehead and the nose extracted from the coronal head image 30, and the vertical line passing through the center of the lower central incisor (LDM) from the oral panoramic image 40, and measures the distance in the Y-axis direction in the coronal head image between the Nasion true vertical plane (NTVP), which is the vertical plane passing through the nasion point (N), and the vertical line passing through the center of the lower central incisor (LDM), thereby deriving the degree of displacement of the center of the lower central incisor, which is one of the 13 parameters, and determining the left-right occlusion relationship between the maxilla and the mandible of the subject. Analysis It can be done.

[0057] In addition, the machine learning algorithm detects the nasion point (N), which is the lowest part in the boundary region between the forehead and the nose extracted from the coronal head image 30, and the lower end point of the right maxillary canine (Ct(Rt)) from the oral panoramic image 40, and the vertical distance between the True horizontal plane (THP), which is the horizontal plane passing through the nasion point, and the lower end point of the right maxillary canine (Ct(Rt)) is derived as one of the 13 parameters, enabling the confirmation of the distance between the horizontal plane and the right maxillary canine.

[0058] In addition, the machine learning algorithm detects the nasion point (N), which is the lowest part in the boundary region between the forehead and the nose extracted from the coronal head image 30, and the lower end point of the left canine (Ct(Lt)) from the oral panoramic image 40, and the vertical distance in the Z-axis direction in the coronal head image 30 between the True horizontal plane (THP), which is the horizontal plane passing through the nasion point, and the lower end point of the left canine (Ct(Lt)) is derived as one of the 13 parameters, enabling the confirmation of the distance between the horizontal plane and the left maxillary canine.

[0059] As a result, the distances between the right and left maxillary canines measured from the horizontal plane should match each other. If there is a difference, the inclination of the maxilla in the canine region can be confirmed.

[0060] In addition, the machine learning algorithm detects the Nasion, which is the lowest part in the boundary region between the forehead and the nose extracted from the coronal head image 30, and the first upper right molar (U6MB(Rt)) from the oral panoramic image 40. The vertical distance between the True horizontal plane (THP), which is the horizontal plane passing through the Nasion, and the first upper right molar (U6MB(Rt)) is derived as one of the 13 parameters, and the distance between the horizontal plane and the first upper right molar can be confirmed.

[0061] In addition, the learning algorithm detects the Nasion (N), which is the lowest part in the boundary region between the forehead and the nose extracted from the coronal head image 30, and the first upper left molar (U6MB(Lt)) from the oral panoramic image 40. The vertical distance between the True horizontal plane (THP), which is the horizontal plane passing through the Nasion, and the first upper left molar (U6MB(Lt)) is derived as one of the 13 parameters, and the distance between the horizontal plane and the first upper left molar can be confirmed.

[0062] As a result, the distances between the first upper right molar and the first upper left molar measured from the horizontal plane should match each other. If there is a difference, the inclination of the maxilla in the molar region can be confirmed.

[0063] In addition, the machine learning algorithm detects the Nasion (N), which is the lowest part in the boundary region between the forehead and the nose extracted from the thalamic head image 30, and the upper endpoint (T1) and the lower endpoint (T2) of the upper anterior teeth on the anterior tooth cross-sectional image 50. The inclination of the upper central incisor is derived as one of the 13 parameters based on the angle between the True horizontal plane (THP), which is the horizontal plane passing through the Nasion, and the vector connecting the upper endpoint (T1) and the lower endpoint (T2) of the upper anterior teeth, and the occlusion state of the subject can be Analysis determined.

[0064] In addition, the machine learning algorithm detects the mental point (Me), which is the lowest point on the mandible, and the gonion (Go), which is the point of maximum curvature on the mandible, from the thalamic head image 30, and the upper end point (T3) and the lower end point (T4) of the mandibular anterior teeth from the anterior tooth cross-sectional image 50, respectively. The inclination of the mandibular central incisor is derived as one of the 13 parameters based on the angle between the MeGo line connecting the mental point (Me) and the gonion (Go) and the vector connecting the upper end point (T3) and the lower end point (T4) of the mandibular anterior teeth, and the occlusal state of the subject is Analysis determined.

[0065] In addition, the machine learning algorithm detects the nasion point (N), which is the lowest part of the boundary region between the forehead and the nose, the mental point (Me), which is the lowest point on the mandible, and the gonion (Go), which is the point of maximum curvature on the mandible, from the thalamic head image 20. The vertical inclination of the mandible with respect to the horizontal reference plane (THP) is derived as one of the 13 parameters based on the angle between the True Horizontal Plane (THP), which is a horizontal plane passing through the nasion point (N), and the MeGo line connecting the mental point (Me) and the gonion (Go), and the vertical jaw relationship of the subject is Analysis determined.

[0066] Figure 4 is a diagram showing the process of detecting a plurality of measurement points for parameter derivation in the thalamic head image using the machine learning algorithm.

[0067] As shown in FIG. 4(a), the machine learning algorithm can divide the subject's thalamic head image 20 into four regions of interest (ROIs) in order to detect a plurality of measurement points. Here, the thalamic head image 20 includes a first region 21 provided between the frontal head and the nostril point located at the lowest position where two nasal bones meet, a second region 23 provided between the nostril point and the upper jaw teeth, a third region 25 provided between the upper jaw teeth and the mental point (Me) of the mandible, and a fourth region 29 provided between the mental point of the mandible and the articulare (Ar) of the temporomandibular joint bone. A plurality of measurement points for parameter derivation can be detected in each of the four regions of interest.

[0068] Therefore, the machine learning algorithm is displayed by a red line along the front of the subject's face in order to divide the thalamic head image 20 into four regions of interest respectively, and the facial profile region 27 composed of the first region 21, the second region 23, and the third region 25, and is displayed by a green line along the subject's mandibular region, and the jaw profile region composed of the fourth region 25 can be divided and extracted. From the thalamic head image 20, for the extraction of the jaw profile region composed of the facial profile region 27 and the fourth region 29, the thalamic head image 20 may be divided into a plurality of unit pixels that are horizontal or vertical in the y-axis direction respectively.

[0069] As shown in FIG. 4(a), the process of extracting the subject's facial profile region 27 on the thalamic head image 20 may be performed based on the similarity with other adjacent pixels from the viewpoints of depth and characteristics as the skull boundary at an arbitrary point within the thalamic head image 20. Specifically, the coordinate value D(xi, i) of the unit pixel that is not zero and has the maximum x-axis value on the thalamic head image 20 may be regarded as existing in the facial profile region 27 when the following equation is satisfied when compared with the unit pixel located in the previous row (i - 1).

[0070]

Equation

[0071] Next, in order to extract the jaw profile region including the fourth region 29 based on the facial profile region 27 obtained from the thalamic plane head image 20, the menton (Me) is designated as the starting point. As shown in FIG. 4(b), after dividing the jaw profile region composed of the fourth region 29 in the thalamic plane head image into a plurality of regions of interest 29s, the average depth level in each of the plurality of regions of interest 29s is calculated, and the articular jaw bone (Ar) can be detected by measuring the actual distance between the depth level in the region of interest where the depth level changes suddenly and the vertical plane of the skull.

[0072] Referring to FIGS. 3 and 4, the machine learning algorithm can detect the point at the lowest position along the x-axis direction in the first region 21 constituting the thalamic plane direction head image 20 as the nasion point (N). By recognizing the vertical plane extending along the z-axis direction while passing through the detected nasion point (N) as the Nasion true vertical plane (NTVP), and the horizontal line extending along the y-axis direction while passing through the nasion point (N) as the True horizontal plane (THP), parameters can be derived.

[0073] In addition, the machine learning algorithm can detect the point at the lowest part of the upper anterior teeth in the second region 23 constituting the thalamic plane head image 20 as the point A (A). On the other hand, when it is difficult for the machine learning algorithm to recognize the shape of the upper anterior teeth in the second region 23 constituting the thalamic plane head image, the shape of the upper anterior teeth can be complemented by gently connecting the boundary region between the anterior nasal spine point (acanthion) provided at the upper part of the upper anterior teeth and protruding forward in front of the teeth and the upper anterior teeth. Thereafter, the machine learning algorithm can detect the point at the lowest position in the x-axis direction within the facial profile boundary region or the point with the smallest gradient within the facial profile boundary region as the point A.

[0074] Further, the machine learning algorithm can detect a point B (B) which is the lowest point in the x-axis direction from the mandible, a pogonion (Pog) which is the highest point in the x-axis direction from the mandible, and a menton (Me) which is the lowest point in the y-axis direction from the mandible in a third region 25 constituting the thalamic head image 20, respectively.

[0075] On the other hand, when the mandible of the subject is located more inward than the anterior mandibular teeth, the point B (B) and the pogonion (Pog) cannot be smoothly detected by the above method. In this case, the machine learning algorithm can detect the most concave point and the bulging point on the mandible in the third region 25 constituting the thalamic head image 20 as the point B (B) and the pogonion (Pog), respectively.

[0076] Further, the machine learning algorithm can detect a gonion (Go) which is the point of maximum curvature of the mandible in a fourth region 29 constituting the thalamic head image 20. For this purpose, the machine learning algorithm can detect the intersection point of a tangent line that contacts the lower part of the mandible while passing through the menton (Me) and a tangent line that contacts the left side of the mandible while passing through the articular bone as the gonion (Go).

[0077] FIG. 5 is a diagram showing a process of detecting a plurality of measurement points in a coronal plane head image using a machine learning algorithm.

[0078] As shown in FIG. 5, the machine learning algorithm can divide the coronal plane head image 30 including two regions of interest (ROIs) in order to detect a plurality of measurement points. The machine learning algorithm divides the coronal plane head image 30 into a fifth region 31 included between the nostril points which are the lowest points where the eyes and the two nasal bones meet in the facial region and a sixth region 33 which is a mandibular region, and can detect head measurement points respectively.

[0079] On the other hand, since the nasal root points (N) detected in the coronal head image 30 and the thalamic head image 20 share the same z-axis position coordinate, the y-axis position coordinate in the coronal head image 30 can act as a key element in the process of detecting the nasal root point (N). From TIFF0007701694000003.tif1426 TIFF0007701694000004.tif1426 (where, Z N is the z-axis position coordinate of point N, where n is a natural number. ) from the multiple unit pixels included in i ) and the right edge coordinate value (T' i ) is detected, the y-axis position coordinate (y N ) may be detected by the following formula 2:

[0080]

number

[0081] In the process of detecting the sixth region 33 from the coronal head image 30, the machine learning algorithm calculates the z-axis position coordinate (Z A ) and the z-axis position coordinate of point B (Z B ) z-axis position coordinate of the center point The intersection points (S1, S2) where a vertical line passing through TIFF0007701694000006.tif1030 meets the left mandible and the right mandible in the coronal head image 30, respectively, are detected, and the lower area of ​​the detected pair of intersection points (S1, S2) can be designated as the sixth area 33 in the coronal head image 30.

[0082] Then, a region that bulges in the z-axis direction is detected from the sixth region 33 of the coronal head image 30, and the point in the region where the x-axis coordinate value is maximum can be detected as Menton (Me).

[0083] Figs. 6 to 10 are diagrams showing the process of detecting a plurality of measurement points in an oral panoramic image using a machine learning algorithm.

[0084] Fig. 6 is a diagram showing the process of applying an R-CNN (Region based convolutional neural networks) machine learning model to an oral panoramic image 40 based on a machine learning algorithm to sense each individual tooth region 41 from the entire teeth of a subject. The individual tooth regions 41 may be shown in a plurality of different colors so as to be distinguished from adjacent regions.

[0085] Fig. 7 is a diagram showing the process of detecting a tooth measurement point 42 indicating the position of each tooth for each individual tooth region 41 in the sensed entire teeth. Here, the tooth measurement point 42 may be detected as the central point inside each individual tooth region 41 in the entire teeth detected from the Mask R-CNN model.

[0086] Fig. 8 is a diagram showing the process of analyzing the positions of the detected tooth measurement points 42 and classifying the entire teeth of the subject into upper teeth 40a and lower teeth 40b in the oral panoramic image 40.

[0087] As a statistical method for the classification, by the linear regression method, two-dimensional position coordinates corresponding to the positions of the plurality of detected tooth measurement points 42 are set respectively, and a quadratic function 43 passing through the coordinates can be generated. The relative positions of the positions of the tooth measurement points 42 sensed in Fig. 7 and the quadratic function 43 are sensed, and the tooth measurement points 42 are classified into upper tooth measurement points 42a and lower tooth measurement points 42b, whereby the entire teeth appearing on the oral panoramic image 40 can be classified into upper teeth 40a or lower teeth 40b.

[0088] FIG. 9 is a diagram showing a process of numbering right upper teeth 40a1, left upper teeth 40a2, right lower teeth 40b1, and left lower teeth 40b2, respectively, by calculating the distance from each of the upper teeth 40a and lower teeth 40b to a tooth measurement point 42 detected based on the horizontal distance from the facial midline 44 on the oral panoramic image 40.

[0089] By the numbering process, all the teeth of the subject appearing on the oral panoramic image 40 may be sequentially numbered in ascending order of the horizontal distance from the facial midline 44 to the detected measurement points (42, see FIG. 8).

[0090] Also, by the numbering process, an abnormal deviation in the distance from the facial midline 44 to the tooth measurement points detected from each of two adjacent teeth can be sensed, and a missing tooth 40m can be sensed. In the embodiment of FIG. 9, the machine learning algorithm can confirm that the first molar (tooth No. 6) of the right upper teeth 40a1 is the missing tooth 40m.

[0091] FIG. 10 is a diagram showing a process of detecting a plurality of measurement points for deriving parameters from detected target teeth including incisors, canines, and first molars among all the teeth appearing on the oral panoramic image 40 by analyzing the numbered teeth.

[0092] Referring to FIG. 9, the detected target teeth 45 for which detection of measurement points is required for deriving the 13 parameters are the incisors (tooth No. 1), canines (tooth No. 3), first molars (tooth No. 6) in the right upper teeth 40a1, the incisors (tooth No. 1), canines (tooth No. 3), first molars (tooth No. 6) in the left upper teeth 40a2, the incisors (tooth No. 1) in the right lower teeth 40b1, and the incisors (tooth No. 1) in the left lower teeth 40a2 among all the teeth appearing on the oral panoramic image 40.

[0093] As shown in FIG. 10, the machine learning algorithm adjusts the size of a region of interest (ROI) including a plurality of teeth 45 to be detected, including the front teeth, canine teeth, and first molars described above, and loads the region of interest for the adjusted teeth to be detected into the CNN model, whereby three tooth measurement points 47 can be detected from each of the plurality of teeth 45 to be detected.

[0094] Here, the three tooth measurement points 47 detected from each of the plurality of teeth 45 to be detected are composed of a left tooth measurement point (P1), a central tooth measurement point (P2), and a right tooth measurement point (P3) at the tooth enamel part of the tooth crown constituting the tooth. As a result, based on the position coordinates 48 for each of the three tooth measurement points detected from the teeth 45 to be detected learned from the CNN model, a plurality of measurement points for deriving parameters for each of the teeth 45 to be detected may be detected on the oral panoramic image 40.

[0095] For example, in order to detect a plurality of measurement points for deriving parameters, the machine learning algorithm can define the central point of the central tooth measurement point (P2) among the three measurement points detected for each of the upper front teeth and the lower front teeth in the oral parameter image 40 shown in FIG. 10 as the upper central incisor center point (UDM) and the lower central incisor center point (LDM).

[0096] On the other hand, three tooth measurement points (P1, P2, P3) are detected from each of the four front teeth 45 sensed by the oral panoramic image 40, and based on the three tooth measurement points (P1, P2, P3) detected for each of the four front teeth 45, a front tooth cross-sectional image 50 may be acquired from the oral panoramic image 40 (see FIG. 3). Referring to FIG. 3, the machine learning algorithm can accurately detect the upper front tooth upper endpoint (T1), the upper front tooth lower endpoint (T2), the lower front tooth upper endpoint (T3), and the lower front tooth lower endpoint (T4) on the front tooth cross-sectional image 50 by utilizing CNN model distributed processing.

[0097] The machine learning algorithm measures the angle between the vector connecting the upper anterior tooth upper endpoint (T1) and the upper anterior tooth lower endpoint (T2) detected on the anterior tooth cross-sectional image and the horizontal plane (NTVP) passing through the N points, and measures the angle between the vector connecting the lower anterior tooth upper endpoint (T3) and the lower anterior tooth lower endpoint (T4) and the MeGo line. As a result, the anterior tooth inclination is evaluated and can be used as an orthodontic diagnosis parameter.

[0098] As described above, the machine learning-based orthodontic head measurement parameter derivation method according to the present invention corresponds to the 13 parameters derived in the 13 parameter derivation steps (S300), and the facial shape or occlusion state of the subject is Analysis It may further include a step (S400) of doing.

[0099] Here, when the occlusion state of the subject is Analysis performed, in order to distinguish normal occlusion from malocclusion, the machine learning algorithm uses the internally stored parameter reference values to determine the state in which the anteroposterior relationship between the maxilla and the mandible is in a relatively normal range, the state in which the maxilla protrudes more than the mandible, and the state in which the mandible protrudes more than the maxilla. Each can be classified Analysis and done.

[0100] In addition, when the facial shape of the subject is Analysis performed corresponding to the 13 derived parameters, the machine learning algorithm analyzes the degree of the facial length of the subject to perform a correction diagnosis. According to the internally stored parameter reference values, the state in which the facial length is in the normal range, the state in which the facial length is shorter than the normal range, and the state in which the facial length is longer than the normal range are respectively classified Analysis and done.

[0101] FIG. 11 is a diagram showing a screen of a graphic user interface to which the machine learning-based head measurement parameter derivation method according to the present invention is applied, and FIG. 12 is a diagram showing an enlarged view of a diagnosis result display area on the screen of the graphic user interface shown in FIG. 11.

[0102] In the graphic user interface (GUI) to which the machine learning-based head measurement parameter derivation method shown in FIG. 11 is applied, each step constituting the machine learning-based head measurement parameter derivation method is executed through a display, and a plurality of images or input / output icons showing the process of executing each step may be displayed on the screen of the display.

[0103] For example, in the machine learning-based head measurement parameter derivation method, when the step (S100) of acquiring a diagnostic head measurement image is executed, a plurality of diagnostic head measurement images including video data 10 obtained by three-dimensional cone beam computed tomography (CBCT) of the face of the subject, thalamic head image 20 extracted from the CBCT video data 10, coronal head image 30, oral panoramic image 40, and anterior tooth cross-sectional image 50 may be displayed on the display screen through the head measurement image generation area 100.

[0104] Then, in the machine learning-based head measurement parameter derivation method, when the step (S200) of detecting a plurality of measurement points is executed, measurement points for deriving parameters by the machine learning algorithm of the present invention are automatically detected from the measurement point display area 200, or icons such as various symbols and figures corresponding to the detected measurement points may be displayed on the display screen so that a skilled person such as a dental specialist can directly display the measurement points manually detected on the display screen.

[0105] As shown in FIG. 12, in the machine learning-based head measurement parameter derivation method, after the step (S300) of deriving 13 parameters by the machine learning algorithm input to the processor constituting the user graphic interface (GUI) device is performed when the step (S200) of detecting a plurality of measurement points is performed, information 310 regarding the 13 derived parameters may be displayed through the diagnostic result display area 300 displayed on the display screen.

[0106] On the one hand, in the method for deriving head measurement parameters based on machine learning according to the present invention, the facial shape or occlusion state of the subject is Analysis performed in step (S400) corresponding to the 13 parameters derived in the 13 parameter derivation steps (S300), as described above.

[0107] Therefore, in the method for deriving orthodontic treatment based on machine learning, when the step (S400) of Analysis performing the facial shape or occlusion state of the subject is executed, Analysis through the result display area 300, information 320 automatically corresponding to the 13 parameters for the occlusion state or facial shape of the subject may be displayed on the display screen. Analysis For example, when the anteroposterior occlusion state between the maxilla and the mandible of the subject is in a relatively normal range, the diagnostic information is "Class I", when the maxilla of the subject protrudes more than the mandible, it is "Class II", and when the mandible protrudes more than the maxilla, it can be indicated by words such as "Class III".

[0108] Also, information 320 regarding the facial shape of the subject can be displayed in the diagnostic result display area 300 of the display screen, such as words like "Meso - cephalic facial pattern" for the state where the length of the subject's facial part is in the normal range, "Brachy - cephalic facial pattern" for the state where the length of the facial part is shorter than the normal range, and "Dolicho - cephalic facial pattern" for the state where the length of the facial part is longer than the normal range.

[0109]

[0110] ​Such a method for deriving head measurement parameters is programmed into a head measurement parameter derivation program and may be installed or stored in a user computing device or a computable cloud server. Such a program uses, as input data, the head diagnostic measurement image extracted at the stage of obtaining the diagnostic head measurement image of the head measurement parameter derivation method described above, and detects the plurality of measurement points as output data; and the derivation program may be programmed such that the stage of deriving 13 parameters corresponding to the distances or angles between the plurality of detected measurement points is automatically performed.

[0111] Of course, the stage of detecting the plurality of measurement points as output data and the stage in which the derivation program derives 13 parameters corresponding to the distances or angles between the plurality of detected measurement points may be programmed to be performed sequentially or stepwise according to the user's selection.

[0112] Thus, in the machine learning-based head measurement parameter derivation method according to the present invention, from the video data obtained by CBCT imaging of a subject, a diagnostic head measurement image extracted at a specific angle by applying a machine learning algorithm is obtained, and the entire process for deriving 13 parameters corresponding to a plurality of measurement points detected from the diagnostic head measurement image is performed in several seconds to several tens of seconds. Therefore, the derivation of parameters for orthodontic diagnosis can be performed quickly, accurately, and consistently.

[0113] In addition, when the machine learning-based head measurement parameter derivation method of the present invention is combined with a graphical user interface, by displaying the results obtained at each of the respective stages on a display screen, a third party such as the subject and a dental professional can smoothly grasp the measurement points for orthodontic diagnosis, the derivation process of the 13 parameters, and the diagnostic results.

[0114] Note that the machine learning-based head measurement parameter derivation method according to the present invention automatically determines the facial shape or occlusion state of the subject AnalysisFurthermore, advancing from the effect of displaying it, it has an excellent expected effect with a further expanded application range, such as automatically designing a customized dental orthosis for the subject corresponding to the derived 13 parameters.

[0115] This specification has been described with reference to preferred embodiments of the present invention. However, those skilled in the art in the relevant technical field will be able to implement the present invention with various modifications and changes without departing from the spirit and scope of the present invention described in the claims below. Therefore, any modified implementation should be regarded as being included in the technical scope of the present invention when it basically includes the components of the claims of the present invention.

Claims

1. A method for deriving head measurement parameters for machine learning-based orthodontic diagnosis using head measurement images for diagnosis extracted from video data obtained by three-dimensional cone beam computed tomography (CBCT) of a subject's head in a natural head position state, comprising, for the subject, a thalamic head image, a coronal head image, an oral panoramic image, and a front tooth cross-sectional image, detecting a plurality of measurement points for deriving 13 parameters for orthodontic diagnosis on the head measurement image based on a machine learning algorithm; and deriving 13 parameters corresponding to distances or angles between the plurality of detected measurement points, wherein the 13 parameters include parameters derived using the Nasion True Vertical Plane (NTVP), which is a vertical plane passing through the Nasion, which is the lowest part in the boundary region between the forehead and the nose, among the plurality of measurement points, or the True Horizontal Plane (THP), which is a horizontal plane passing through the Nasion. A method for deriving head measurement parameters for orthodontic diagnosis, characterized by the above.

2. The 13 parameters are, in order to provide information for orthodontic diagnosis, the protrusion of the maxilla, the protrusion of the mandible, the protrusion of the chin tip, the degree of displacement of the center of the mandible, the degree of displacement of the center of the maxillary central incisor, the degree of displacement of the center of the mandibular central incisor, the vertical distance from the True Horizontal Plane (THP) passing through the Nasion, which is the lowest part in the boundary region between the forehead and the nose, to the lower end point of the right canine tooth, the vertical distance from the THP to the lower end point of the left canine tooth, the vertical distance from the THP to the right maxillary first molar, the vertical distance from the THP to the left maxillary first molar, the inclination of the maxillary central incisor, the inclination of the mandibular central incisor, and the vertical inclination of the mandible with respect to the THP. The method for deriving head measurement parameters for orthodontic diagnosis according to Claim 1, characterized by the above.

3. The machine learning algorithm divides the area provided between the frontal part and the rhinion on the thalamic head image, the area from the rhinion to the upper teeth, the area from the upper teeth to the Menton of the mandible, and the area from the Menton of the mandible to the articulare of the temporomandibular joint, and detects a plurality of measurement points for deriving the 13 parameters. The method for deriving head measurement parameters for orthodontic diagnosis according to claim 1, characterized in that.

4. The machine learning algorithm divides the area from the frontal part to the rhinion on the coronal head image and the mandibular area, and detects a plurality of measurement points for deriving the 13 parameters. The method for deriving head measurement parameters for orthodontic diagnosis according to claim 1, characterized in that.

5. The machine learning algorithm includes a process of applying an R-CNN (Region based convolutional neural networks) machine learning model to an oral panoramic image to sense individual areas of all teeth; A process of detecting tooth measurement points indicating the positions of teeth for each of the individual areas of the sensed all teeth; A process of classifying all teeth into upper teeth and lower teeth by analyzing the positions of the detected tooth measurement points; A process of sequentially numbering the right upper teeth, left upper teeth, right lower teeth, and left lower teeth according to the horizontal distance from the facial midline to the detected tooth measurement points; and A process of analyzing the numbered teeth and detecting a plurality of measurement points for deriving parameters in specific teeth including incisors, canines, and first molars; The method for deriving head measurement parameters for orthodontic diagnosis according to claim 1, characterized by including.

6. In the thalamic head image, the machine learning algorithm detects the A point (A point, A), which is the deepest part on the line connecting the Nasion, which is the lowest part in the boundary area between the forehead and the nose, the Anterior nasal spine in the maxilla, and the Prosthion of the maxillary anterior tooth alveolus. The method for deriving head measurement parameters for orthodontic diagnosis according to claim 2, characterized in that the protrusion of the maxilla is derived by measuring the distance between the Nasion true vertical plane (NTVP), which is a vertical plane passing through the Nasion, and the A point.

7. The machine learning algorithm detects, in the thalamic head image, the Nasion, which is the lowest part in the boundary region between the forehead and the nose, the Infradentale, which is the lowest part in the alveolar region of the lower anterior teeth, and the B point (B point, B), which is the deepest part connecting the Nasion, the Infradentale, and the Pog (Pog). The method for deriving a head measurement parameter for orthodontic diagnosis according to claim 2, wherein the protrusion of the mandible is derived by measuring the distance between the Nasion true vertical plane (NTVP), which is a vertical plane passing through the Nasion, and the B point.

8. The machine learning algorithm detects, in the thalamic head image, the Nasion, which is the lowest part in the boundary region between the forehead and the nose, and the Pogonion (Pogonion, Pog), which protrudes most forward from the mandible. The method for deriving a head measurement parameter for orthodontic diagnosis according to claim 2, wherein the protrusion of the pogonion is derived by measuring the distance between the Nasion true vertical plane (NTVP), which is a vertical plane passing through the Nasion, and the Pogonion.

9. The machine learning algorithm detects, in the coronal head image, the Nasion, which is the lowest part in the boundary region between the forehead and the nose, and the Menton, which is the lowest point on the mandible. The method for deriving a head measurement parameter for orthodontic diagnosis according to claim 2, wherein the degree of displacement of the center of the mandible is derived by measuring the distance between the Nasion true vertical plane (NTVP), which is a vertical plane passing through the Nasion, and the Menton.

10. The machine learning algorithm detects, in the coronal head image, the Nasion, which is the lowest part in the boundary region between the forehead and the nose, and the central point of the upper central incisor in the oral panoramic image. The method for deriving a head measurement parameter for orthodontic diagnosis according to claim 2, wherein the degree of displacement of the center of the upper central incisor is derived by measuring the distance between the Nasion true vertical plane (NTVP), which is a vertical plane passing through the Nasion, and the central point of the upper central incisor.

11. The machine learning algorithm detects, in the coronal head image, the Nasion, which is the lowest part in the boundary region between the forehead and the nose, and the central point of the lower central incisor on the oral panoramic image. The method for deriving head measurement parameters for orthodontic diagnosis according to claim 2, characterized in that the degree of displacement of the center of the mandibular central incisor is derived by measuring the distance between the Nasion true vertical plane (NTVP), which is a vertical plane passing through the nasion point, and the center point of the mandibular central incisor.

12. The machine learning algorithm detects the nasion point, which is the lowest part in the boundary region between the forehead and the nose in the coronal head image, and the lower end point of the right canine tooth on the oral panoramic image. The method for deriving head measurement parameters for orthodontic diagnosis according to claim 2, characterized in that the vertical distance between the True horizontal plane (THP), which is a horizontal plane passing through the nasion point, and the lower end point of the right canine tooth is derived.

13. The machine learning algorithm detects the nasion point, which is the lowest part in the boundary region between the forehead and the nose in the coronal head image, and the lower end point of the left canine tooth in the oral panoramic image. The method for deriving head measurement parameters for orthodontic diagnosis according to claim 2, characterized in that the vertical distance between the True horizontal plane (THP), which is a horizontal plane passing through the nasion point, and the lower end point of the left canine tooth is derived.

14. The machine learning algorithm detects the nasion point, which is the lowest part in the boundary region between the forehead and the nose in the coronal head image, and the lower end point of the right first maxillary molar on the oral panoramic image. The method for deriving head measurement parameters for orthodontic diagnosis according to claim 2, characterized in that the vertical distance from the THP to the right first maxillary molar is derived by the distance between the True horizontal plane (THP), which is a horizontal plane passing through the nasion point, and the lower end point of the right first maxillary molar.

15. The machine learning algorithm detects the nasion point, which is the lowest part in the boundary region between the forehead and the nose in the coronal head image, and the lower end point of the left first maxillary molar on the oral panoramic image. The method for deriving head measurement parameters for orthodontic diagnosis according to claim 2, characterized in that the vertical distance from the THP to the left first maxillary molar is derived by the distance between the True horizontal plane (THP), which is a horizontal line passing through the nasion point, and the lower end point of the left first maxillary molar.

16. The machine learning algorithm detects the nasion, which is the lowest part in the boundary region between the forehead and the nose in the coronal plane head image, and the upper and lower endpoints of the maxillary anterior teeth on the anterior dental cross-sectional image. The method for deriving head measurement parameters for orthodontic diagnosis according to claim 2, wherein the inclination of the maxillary central incisor is derived based on the angle of the vector connecting the THP (True horizontal plane), which is the horizontal plane passing through the nasion, and the upper and lower endpoints of the maxillary anterior teeth.

17. The machine learning algorithm detects the menton, which is the lowest point of the mandible, the gonion, which is the point of maximum curvature of the mandible, in the thalamic plane head image, and the upper and lower endpoints of the mandibular anterior teeth in the anterior dental cross-sectional image. The method for deriving head measurement parameters for orthodontic diagnosis according to claim 2, wherein the inclination of the mandibular central incisor is derived based on the angle between the MeGo line connecting the menton and the gonion and the vector connecting the upper and lower endpoints of the mandibular anterior teeth.

18. The machine learning algorithm detects the nasion, which is the lowest part in the boundary region between the forehead and the nose in the thalamic plane head image, the menton, which is the lowest point of the mandible, and the gonion, which is the point of maximum curvature of the mandible. The method for deriving head measurement parameters for orthodontic diagnosis according to claim 2, wherein the vertical inclination of the mandible with respect to the THP is derived based on the angle between the THP (True horizontal plane) passing through the nasion and the MeGo line connecting the menton and the gonion.

19. The method for deriving head measurement parameters for orthodontic diagnosis according to claim 1, further comprising analyzing the facial shape or occlusion state of the subject corresponding to the 13 derived parameters.

20. When analyzing the occlusion state of the subject corresponding to the 13 derived parameters, the method for deriving head measurement parameters for orthodontic diagnosis according to claim 19 is characterized in that the anteroposterior occlusion state between the maxilla and the mandible in a relatively normal range, the state where the maxilla protrudes more than the mandible, and the state where the mandible protrudes more than the maxilla are classified and analyzed respectively.

21. When analyzing the facial shape of a subject corresponding to the derived 13 parameters, the method for deriving head measurement parameters for orthodontic diagnosis according to claim 19, characterized in that the state where the length of the facial part is within the normal range, the state where the length of the facial part is shorter than the normal range, and the state where the length of the facial part is longer than the normal range are classified and analyzed respectively.

22. A step of using the head diagnostic head measurement image extracted at the stage of obtaining the diagnostic head measurement image of the method for deriving head measurement parameters for orthodontic diagnosis according to any one of claims 1 to 21 as input data and detecting the plurality of measurement points as output data; and, A step of deriving 13 parameters corresponding to the distances or angles between the plurality of detected measurement points; being programmed to be automatically performed and installed in a computing device or a computable cloud server, a head measurement parameter derivation program for orthodontic diagnosis.

Images