Method, device, and computer-readable recording medium for separating dental objects from 3D oral scan data, automatically detecting tooth malpositions, and recommending orthodontic treatment plans

Abstract

The present invention relates to a method for separating tooth objects from 3D oral scan data, automatically detecting abnormal tooth position, and recommending an orthodontic treatment plan, and specifically, includes a first contact point search step for sequentially increasing the size of a circle at the center of simple coordinates of 3D oral scan data corresponding to pre-treatment oral state information of a subject, and searching for a first point where the circle and an object included in the 3D oral scan data contact each other; and a sphere placement step for distributing one or more spheres at a predetermined interval within the tooth at the first contact point searched, reversing the vector direction of the mesh, and duplicating the spheres while moving the spheres in a circumferential direction at a random speed in the reference coordinates, and placing the spheres within the tooth. a feature data extraction step of growing the size of the replicated spheres to a predetermined size to determine representative spheres corresponding to each tooth and extracting feature data of the teeth based on coordinates and contact points of the representative spheres; an image processing step of expressing the subject's tooth arrangement having an individually separated state in a 3D image using the feature data extracted for each tooth and the tooth-specific axis data; a positional abnormality classification step of classifying positional abnormalities by applying the relative arrangement state of the tooth objects included in the tooth arrangement data to an automatic diagnosis algorithm; and a correction data generation step of combining the classification result for tooth positional abnormality and generating orthodontic data based on the subject's pre-treatment oral cavity condition information.

Description

[Technical Field]

[0001] The present invention relates to a method for separating tooth objects from 3D oral scan data, automatically detecting tooth positional abnormalities, and recommending an orthodontic treatment plan. Specifically, the present invention relates to a technology for separating tooth objects from a subject's 3D oral scan data based on the contact points and feature points of individual teeth, automatically detecting tooth positional abnormalities required for establishing an orthodontic treatment plan using the axial relationships along with accurate relative coordinates and direction values ​​for each tooth, and selecting an optimal tooth arrangement structure, thereby providing an orthodontic treatment plan that reflects the characteristics of the subject's individual teeth. [Background technology]

[0002] In general, in the field of orthodontic treatment in dentistry, various attempts have been made to use image processing techniques to extract meaningful data for orthodontics and to use this data in orthodontic treatment.

[0003] However, in the field of orthodontic treatment, the method of using image processing is limited to the extent that dentists use 3D oral scan data of the patient. As a result, accurate orthodontic treatment is possible only after dentists have acquired know-how through several years of training. For this reason, orthodontic treatment has the problem of large deviations in orthodontic results depending on the dentist, even for the same patient.

[0004] On the other hand, as a way to reduce the deviation in orthodontic results, prior art such as Patent Document 1 proposes a technology that creates tooth modeling based on 3D dental scan data, and reflects the doctor's examination and diagnosis results on the tooth model, enabling simulated orthodontic treatment.

[0005] However, such conventional techniques merely utilize the dentist's independent diagnostic data or diagnostic data dependent on simple experience as attribute data for deriving orthodontic results and compare changes before and after orthodontic treatment, and are limited in their ability to substantially reduce the deviations in orthodontic results depending on the dentist or dental hospital. In order to resolve the above-mentioned problems, the need for an algorithm that derives a diagnosis and treatment plan based on a standardized treatment protocol is emphasized. [Prior art documents] [Patent documents]

[0006] [Patent Document 1] Korean Patent Publication No. 2016-0004862 Summary of the Invention [Problem to be solved by the invention]

[0007] An object of the present invention is to provide a technique for separating tooth objects from 3D oral cavity scan data based on contact points and feature points of individual teeth.

[0008] Another object of the present invention is to provide a technology that, in the field of orthodontics, efficiently analyzes a patient's dental data and derives the results of an automated orthodontic diagnosis for each tooth, thereby constructing such a model and learning it through machine learning, specifically deep learning, to automatically perform diagnoses that are the basis of orthodontic treatment based on 3D oral scan data.

[0009] Furthermore, another object of the present invention is to provide an algorithm for recommending an orthodontic treatment plan that selects an optimal tooth arrangement structure using a predetermined algorithm based on the dental condition and the possibility of individual tooth movement confirmed from the subject's pre-treatment oral condition information, and reflects the characteristics of individual teeth. [Means for solving the problem]

[0010] To achieve the above object, according to one embodiment of the present invention, a method for separating dental objects from 3D oral scan data, automatically detecting tooth positional abnormalities, and recommending an orthodontic treatment plan, which is implemented in a computing device including one or more processors and a main memory storing instructions executable by the processors, includes: an initial contact point search step of sequentially increasing the size of a circle at the center of the simple coordinates of the 3D oral scan data corresponding to the subject's pre-treatment oral condition information, and searching for an initial point where the circle and an object included in the 3D oral scan data meet; a sphere placement step of distributing one or more spheres within the tooth at a predetermined interval from the first contact point searched, reversing the vector direction of the mesh, and moving the spheres in a peripheral direction at a random speed in the reference coordinates, thereby duplicating the spheres at a predetermined period, so that the spheres are uniformly distributed within the tooth; and The method further comprises: a feature data extraction step of growing the created spheres to a predetermined size to determine representative spheres corresponding to each tooth, and extracting tooth feature data based on the coordinates and contact points of the determined representative spheres; an image processing step of obtaining mesh information for each tooth using the feature data extracted for each tooth and tooth-specific axis data, thereby representing the subject's tooth arrangement in a 3D image, each of which is individually separated; a positional abnormality classification step of automatically classifying a plurality of positional abnormalities, including at least the shape of malocclusion, by applying the relative arrangement state of the tooth objects included in the tooth arrangement data generated in the image processing step to an automatic diagnosis algorithm; and a positional abnormality generation step of combining each tooth image, each tooth data, each tooth identification information, and the classification result for each tooth positional abnormality according to the execution result of the positional abnormality classification step, and generating orthodontic data based on the subject's pre-treatment oral condition information.

[0011] In the image processing step, the tooth-specific axis data is extracted based on the volume center and the width center of the 3D oral cavity scan data, and a numerical correction value derived by statistical analysis is applied to a line passing through the volume center and the width center to extract tooth-specific axis data obtained parallel to the tooth roots.

[0012] The representative sphere is determined by growing the size of the replicated spheres to a predetermined size, dropping off the spheres that have fallen out of the teeth, and then shrinking the spheres remaining in the teeth to a predetermined size so that a representative sphere corresponding to the center point of each tooth is determined, and this is continued until a representative sphere corresponding to the number of teeth of the subject is obtained.

[0013] The image processing step converts and exports the mesh information of the individual teeth into a predetermined file format compatible with a first program that handles motion data, and then saves the working data in the first program as a script file.

[0014] In the feature data extraction step, x, y, and z coordinate values ​​for position in three-dimensional space, x, y, and z coordinate values ​​for rotation, and x, y, and z coordinate values ​​for size are extracted as feature data of the representative sphere corresponding to the individual tooth.

[0015] In the sphere placement step, the size of the spheres dispersed within the tooth is set to have a size range of 1 to 5 units, with 1 unit corresponding to 1 / 10,000 mm.

[0016] The positional abnormality classification step, based on the size and relative arrangement of the teeth, classifies each malocclusion as crowding if the interdental contact points do not meet each other but intersect, gap if there is a space between the interdental contact points, rotation if the teeth are rotated relative to each other, perpendicular relationship if the perpendicular relationship of the teeth does not match a predetermined first value, mesiodistal inclination if the gradient of the mesiodistal axial inclination of the teeth does not match a predetermined second value, buccolingual inclination if the gradient of the buccolingual axial inclination of the teeth does not match a predetermined third value, and interdigitation if the cusp of the maxillary tooth is not located between two mandibular teeth.

[0017] In the positional abnormality classification step, the degree of detail is determined based on the size of the numerical value that is the basis for determining each malocclusion, and is set together with the malocclusion classification information.

[0018] In the orthodontic data generation step, mesh information of the tooth image derived from the 3D oral cavity scan data is converted into object format that can be simulated by the first program, and data is generated together with the pre-treatment oral cavity condition information.

[0019] After the orthodontic data generating step is executed, the method further includes a pre-treatment oral condition information loading step of loading pre-treatment oral condition information generated as a result of executing the orthodontic data generating step; a predicted tooth arrangement model deriving step of deriving one or more targeted predicted tooth arrangement models by considering the predicted direction and amount of movement of individual teeth included in the pre-treatment oral condition information of the subject according to an algorithm with previously learned learning data for tooth arrangement; an orthodontic treatment plan presenting step of selecting an orthodontic solution to realize the predicted tooth arrangement model and then presenting a subdivided orthodontic treatment plan by stages; and a feedback collecting step of selecting one or more professionals who match the pre-treatment oral condition information of the subject from a professional talent pool in which a large number of professionals involved in the field of orthodontic treatment are registered, and collecting feedback on the presented orthodontic treatment plan.

[0020] the orthodontic treatment plan presenting step uses a predicted value of an element including at least one of an orthodontic treatment period, an orthodontic force, and a difficulty level of the orthodontic treatment according to the current dental condition of the subject as an objective function; A Pareto optimal solution calculation method is used to search for a Pareto optimal solution that simultaneously satisfies the objective functions and minimizes the value of the objective functions, thereby presenting multiple orthodontic treatment plans.When multiple orthodontic treatment plans are derived by the Pareto optimal solution calculation method, a candidate algorithm including at least one of ANN (Artificial Neural Network), RF (Random Forest), SVM (Support Vector Machine), and EDN (Evolving Deep Network) is used to calculate the optimal orthodontic treatment plan from the multiple orthodontic treatment plans for the subject's pre-treatment oral condition information.

[0021] In executing the pre-treatment oral condition information loading step, image data including at least one of X-ray images, intraoral images, and facial images for the subject's pre-treatment oral condition information is further collected as reference data, and the reference data is used only as a condition variable for the candidate algorithm.

[0022] The orthodontic treatment plan presentation step processes a time-series tooth movement path to realize the predicted tooth arrangement model using a time-series model based on the calculated optimal orthodontic treatment plan and provides it.

[0023] In the execution of the feedback collection step, the selection of the expert is performed by selecting an expert whose clinical history corresponding to the subject's pre-treatment oral condition information is judged to be above a predetermined critical standard, and in the feedback collection step, the feedback collected from one or more experts is labeled so that supervised learning for the algorithm can be performed.

[0024] Meanwhile, a device for separating dental objects from 3D oral scan data, automatically detecting tooth positional abnormalities, and recommending an orthodontic treatment plan is implemented by a computing device including one or more processors and a main memory for storing instructions executable by the processors. The device includes an initial contact point search unit that sequentially increases the size of a circle at the center of a simple coordinate of the 3D oral scan data corresponding to pre-treatment oral condition information of a subject, and searches for an initial point where the circle and an object included in the 3D oral scan data meet; a sphere placement unit that distributes one or more spheres within the tooth at a predetermined interval from the first contact point searched, reverses the vector direction of the mesh, moves the spheres in a circumferential direction at a random speed in a reference coordinate system, and replicates the spheres at a predetermined period, thereby uniformly arranging the spheres within the tooth; and The present invention is characterized by including: a feature data extraction unit that grows the size of spheres to a predetermined size to determine representative spheres corresponding to each tooth, and extracts feature data of the teeth based on the coordinates and contact points of the determined representative spheres; an image processing unit that obtains mesh information of each tooth using the feature data extracted for each tooth and the tooth-specific axis data, thereby representing the subject's tooth arrangement in a 3D image, each of which is individually separated; a positional abnormality classification unit that applies the relative arrangement state of tooth objects included in the tooth arrangement data generated by the image processing unit to an automatic diagnosis algorithm to automatically classify a plurality of positional abnormalities, including at least the shape of malocclusion; and a correction data generation unit that combines each tooth image, each tooth data, each tooth identification information, and the classification result for each tooth positional abnormality according to the function execution result of the positional abnormality classification unit, and generates orthodontic data based on the subject's pre-treatment oral condition information.

[0025] Also, the computer-readable recording medium stores instructions for causing a computing device to perform the following steps, including: an initial contact point search step of sequentially increasing the size of a circle at the center of simple coordinates of 3D oral cavity scan data corresponding to the subject's pre-treatment oral cavity condition information and searching for an initial point where the circle and an object included in the 3D oral cavity scan data meet; a sphere placement step of distributing one or more spheres within the tooth at a predetermined interval from the first contact point searched, reversing the vector direction of the mesh, and moving the spheres in a circumferential direction at a random speed in the reference coordinates while duplicating the spheres at a predetermined period so that the spheres are uniformly distributed within the tooth; and a sphere placement step of growing the size of the spheres replicated in the sphere placement step by a predetermined size. The method further comprises: a feature data extraction step of determining a representative sphere corresponding to each tooth and extracting tooth feature data based on the coordinates and contact points of the determined representative sphere; an image processing step of obtaining mesh information of each tooth using the feature data extracted for each tooth and tooth-specific axis data, thereby representing the subject's tooth arrangement in a 3D image, each of which is in an individually separated state; a positional abnormality classification step of automatically classifying a plurality of positional abnormalities, including at least the shape of malocclusion, by applying the relative arrangement state of the tooth objects included in the tooth arrangement data generated in the image processing step to an automatic diagnosis algorithm; and a step of generating orthodontic data based on the subject's pre-treatment oral condition information by combining each tooth image, each tooth data, each tooth identification information, and the classification result for each tooth positional abnormality according to the execution result of the positional abnormality classification step. [Effects of the Invention]

[0026] This invention solves the problem of dental professionals having to manually separate teeth and gums from 3D oral scan data using specialized 3D programs. It also makes it possible to separate individual tooth objects from 3D oral scan data by extracting the contact points and features of each tooth, thereby increasing the ease of processing tooth image data.

[0027] Furthermore, unlike conventional methods in which tooth feature data is extracted based on the tooth's external shape, the present invention makes it possible to extract tooth feature data based on coordinate system data detection to which the physical characteristics of the tooth are applied, thereby proposing a new approach to tooth feature data extraction.

[0028] Furthermore, compared to existing technologies, the present invention not only extracts dental data that allows for treatment simulation based on 3D oral scan data, but also applies a classification algorithm to the extracted dental data that can automatically learn from the diagnostic results of experienced dentists, thereby generating pre-treatment oral condition information that allows for the establishment of specialized treatment plans regardless of the dentist's level of expertise.

[0029] Furthermore, the present invention selects a predicted tooth arrangement model using a predetermined algorithm based on the dental condition and the possibility of individual tooth movement confirmed from the subject's pre-treatment oral condition information, and provides an algorithm for recommending an orthodontic treatment plan that reflects the characteristics of individual teeth, thereby standardizing orthodontic diagnosis and orthodontic treatment plans.

[0030] In addition, according to the present invention, by standardizing non-standard diagnostic data and utilizing it as reference data for proposing an orthodontic treatment plan, a structured algorithm is designed based on theories and know-how of orthodontic treatment, and by matching a specialist who will carry out the orthodontic treatment plan derived from the subject's pre-treatment oral condition information, the reproducibility of the recommended orthodontic treatment plan and orthodontic treatment results can be increased. [Brief explanation of the drawings]

[0031] [Figure 1] FIG. 1 is a flowchart of a method for separating dental objects from 3D intraoral scan data, automatically detecting dental malpositions, and recommending an orthodontic treatment plan according to one embodiment of the present invention. [Figure 2]FIG. 2 is a diagram illustrating an example of a flow for extracting representative coordinates and contact points of a tooth for generating tooth data according to an embodiment of the present invention. [Figure 3] FIG. 3 is a diagram showing an example of feature data extracted for each individual tooth and embodied in a table according to an embodiment of the present invention. [Figure 4] FIG. 4 is a diagram illustrating a flow of extracting information about a tooth axis according to an embodiment of the present invention. [Figure 5ABCDEFG] 5A to 5G are diagrams showing examples of classification of tooth position abnormalities based on tooth data according to an embodiment of the present invention. [Figure 6AB] FIG. 6AB is a diagram showing an example in which a teeth arrangement model corresponding to the pre-treatment oral cavity condition information of a subject is implemented based on the loaded pre-treatment oral cavity condition information according to an embodiment of the present invention. [Figure 7AB] 7A and 7B are diagrams illustrating an example in which a predicted tooth arrangement model is derived for pre-treatment oral cavity condition information of a subject according to an embodiment of the present invention. [Figure 8] FIG. 8 illustrates an example of an instantiation scheme for an algorithm used to provide a solution for orthodontic treatment according to one embodiment of the present invention. [Figure 9] FIG. 9 illustrates an example problem-solving approach for candidate algorithms from which an optimal orthodontic treatment plan is derived, according to one embodiment of the present invention. [Figure 10] FIG. 10 is a diagram illustrating an example of an interface provided by deriving an optimal orthodontic treatment plan in one embodiment of the present invention. [Figure 11] FIG. 11 is a configuration diagram of an apparatus for separating tooth objects from 3D oral cavity scan data, automatically detecting tooth positional anomalies, and recommending an orthodontic treatment plan according to one embodiment of the present invention. [Figure 12] FIG. 12 is a diagram showing an example of the internal configuration of a computing device according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0032] Various embodiments and / or aspects are described below with reference to the drawings. In the following description, for purposes of explanation, numerous specific details are set forth in order to facilitate a general understanding of one or more aspects. However, those skilled in the art will recognize that these aspects may be practiced without such specific details. The following description and the accompanying drawings set forth in detail certain exemplary aspects of one or more aspects. However, such aspects are illustrative, and only a portion of various methods may be utilized in accordance with the principles of the various aspects, and the description is intended to include all such aspects and their equivalents.

[0033] As used herein, "embodiment," "example," "mode," "exemplary," and the like may not be construed as constituting any described mode or design as better or advantageous over other modes or designs.

[0034] Additionally, the terms "comprise" and / or "comprising" should be understood to mean that the feature and / or component in question is present, but not to exclude the presence or addition of one or more other features, components and / or groups thereof.

[0035] Furthermore, terms including ordinal numbers, such as "first," "second," etc., are used to describe various elements, but the elements are not limited by these terms. These terms are used only to distinguish one element from another. For example, a first element can be referred to as a "second element," and similarly, a second element can be referred to as a "first element," without departing from the scope of the present invention. The term "and / or" includes a combination of multiple related listed items or any of multiple related listed items.

[0036] Furthermore, in the embodiments of the present invention, unless otherwise defined, all terms used herein, including technical and scientific terms, have the same meaning as commonly understood by a person of ordinary skill in the art to which the present invention belongs. Terms defined in commonly used dictionaries should be interpreted as having a meaning consistent with the meaning they have in the context of the relevant art, and should not be interpreted as idealized or overly formal unless explicitly defined in the embodiments of the present invention.

[0037] The present invention relates to a method for separating tooth objects from 3D oral scan data, automatically detecting tooth positional abnormalities, and recommending an orthodontic treatment plan. The first objective of the present invention is to provide a technology for separating tooth objects from 3D oral scan data based on the contact points and feature points of individual teeth. The second objective of the present invention is to provide a technology for automatically performing diagnoses that are fundamental to treatment in orthodontics based on 3D oral scan data by efficiently analyzing a patient's dental data and deriving the results of automatically performing orthodontic diagnoses for each tooth, through the construction of such models and machine learning, specifically, through deep learning model learning. The third objective of the present invention is to provide an algorithm for recommending an orthodontic treatment plan that reflects the characteristics of each tooth by selecting the optimal tooth arrangement structure using a predetermined algorithm based on the dental condition and the possibility of individual tooth movement confirmed from the subject's pre-treatment oral condition information.

[0038] Meanwhile, the following description will be given in more detail with reference to the accompanying drawings, and multiple drawings may be referenced simultaneously to explain one or more technical features or components constituting the invention.

[0039] First, as shown in FIG. 1, in the present invention, an initial contact point search step (S10) is performed in which the size of a circle is sequentially increased at the center of the simple coordinates of the 3D oral scan data corresponding to the subject's pre-treatment oral condition information, and the initial point where the circle and an object included in the 3D oral scan data meet is searched for.

[0040] The 3D oral cavity scan data is data in the form of an STL object, such as a 3D oral cavity scan image typically obtained from a head scanner in the dental field. The STL file format is understood to be a standard data transmission format in the rapid prototyping industry. In addition to the STL file format, the present invention also uses various types of 3D oral cavity scan image formats obtained from 3D scanners in the dental field.

[0041] As a specific example of step S10, referring to FIG. 2, S1 in FIG. 2 shows an example 101 in which a circle is placed at the center of the simple coordinates of the 3D scan data of the subject, and S2 in FIG. 2 shows an example in which the size of the circle placed at the center of the simple coordinates is increased 111 to search for the initial contact point with the tooth object.

[0042] That is, in step S10, the size of the circle is increased at the center of the simple coordinates of the 3D oral cavity scan data to search for the first point that contacts the subject's tooth object, thereby making it possible to easily and conveniently grasp the position of the tooth object on the 3D oral cavity scan data.

[0043] Returning to Figure 1 again, after executing step S10, one or more spheres are dispersed within the tooth at a predetermined interval at the first contact point found in step S10, and then a sphere placement step (S20) is performed in which the vector direction of the mesh information is reversed, and the spheres are moved at a random speed toward the periphery in the reference coordinates, while duplicating the spheres at a predetermined period so that the spheres are uniformly placed within the tooth.

[0044] Here, in the aforementioned step S20, a number of spheres are dispersed at regular intervals at the initial contact points found in step S10 so that the spheres are positioned within the tooth. The size of the dispersed spheres ranges from 1 to 5 units, where 1 unit corresponds to 1 / 10,000 mm, and is understood to be the size of the sphere derived as a result of empirical tuning to find the size of the sphere suitable for the 3D tooth scan data.

[0045] On the other hand, STL files generally have the characteristic of representing the surface of a three-dimensional object as countless triangular faces in three dimensions. Traditionally, the STL file format has had the problem that, while it is easy to detect events of collision with the surface of a tooth object in 3D scan data, it is difficult to detect events of escape from the inside of the tooth object to the outside.

[0046] To solve this problem, the present invention performs a process of regenerating triangular data in the reverse direction by inverting the direction vector defined on the surface of each mesh information consisting of triangles. A specific example of this process will be described later with reference to FIG. 2.

[0047] In detail, S3 in FIG. 2 shows the tooth object before the direction of the mesh information (T) is reversed, and S4 in FIG. 3 shows the tooth object (UT) after the direction of the mesh information (T) is reversed.

[0048] Furthermore, S5 in Figure 2 shows an example in which, due to the inversion process of such mesh information, a large number of spheres are present dispersed within the tooth, and a large number of spheres 121 move at random speeds and are replicated at predetermined intervals. By repeating this process of replicating the spheres 121, it is possible to achieve an even distribution of the spheres 121 within the tooth.

[0049] On the other hand, after the execution of the aforementioned step S20, the size of the spheres duplicated in step S20 is grown to a predetermined size so that a representative sphere corresponding to each tooth is determined, and a feature data extraction step (S30) is performed to extract feature data of the tooth based on the coordinates and contact points of the determined representative sphere.

[0050] In step S30, the size of the spheres dispersed within the tooth object is successively increased, and the spheres that escape from the space within the tooth are dropped out. After that, the spheres remaining within the tooth are reduced to a predetermined size again, so that a representative sphere corresponding to the center point of each tooth is determined.

[0051] Here, it is desirable that the size growth of the spheres mentioned above occurs until the number of spheres remaining in the teeth corresponds to the number of teeth of the subject. For example, if the subject has 14 teeth in the lower jaw, the size growth of the spheres continues until 14 spheres remain, which corresponds to the number of teeth in the lower jaw.

[0052] Furthermore, it is desirable to determine the representative sphere by such growth and shrinkage of the sphere size for both the mandible and the maxilla, but this is understood to reflect the fact that the number of teeth in the maxilla and mandible differs depending on the oral structure of the subject, including the state of tooth extraction.

[0053] As a specific example, as shown in S6 of Figure 2, an example is shown in which a representative sphere 131 is determined for each tooth contained in the subject's lower jaw, and in the present invention, feature data can be extracted using the coordinate values ​​of the representative sphere 131 determined for each individual tooth of the subject.

[0054] Here, it is preferable that the coordinate values ​​of the representative sphere 131 mentioned above are extracted as coordinate values ​​on the x, y, and z planes for the position, rotation, and scale in the three-dimensional space.

[0055] As an example, 200 in Figure 3 shows an example in which feature data of individual teeth is extracted by obtaining coordinate information on the x, y, and z planes for position, rotation, and size from a representative sphere determined for each tooth.

[0056] Although 200 in FIG. 3 shows only feature data for seven individual teeth, this is not limited to this, and it is understood that the tooth feature data referred to in the present invention is extracted in a number corresponding to the number of teeth of the subject.

[0057] Next, after performing the above-mentioned step S30, an image processing step (S40) is performed in which mesh information of individual teeth is obtained using the feature data extracted for each tooth and the tooth-specific axis data, thereby realizing the subject's tooth arrangement in an individually separated state as a 3D image.

[0058] Here, the tooth-specific axis data is extracted based on the volume center point and the width center point of the 3D oral cavity scan data.

[0059] Specifically, the tooth-specific axial data is extracted by applying a numerical correction value derived by statistical analysis to a line passing through the volume center point and the width center point, and axial data obtained parallel to the tooth root is extracted as tooth-specific axial data.

[0060] As an example, at 600 in FIG. 4, it can be seen that the volume center (VC) and area center (AC) points for the individual tooth objects are shown.

[0061] Here, in the present invention, the axis data for each tooth can be extracted as the axis data for each individual tooth by applying a correction value derived by statistical analysis to the line passing through the volume center (VC) and the width center (AC) in Figure 4 to derive an axis parallel to the root of the tooth. The correction value is a value obtained by continuously performing error learning on the axis data for each individual tooth using a learning module installed in a computing device. Such correction values ​​can be the same value or can be derived as different values ​​depending on the shape of the individual tooth.

[0062] Meanwhile, in step S40 of FIG. 1 described above, it is desirable to convert the mesh information of the individual teeth into a predetermined file format compatible with the first program that handles motion data, export it, and then save the working data in the first program as a script file.

[0063] Here, it is understood that the first program is preferably a Unity simulation, and that the predetermined file format is converted into an obj file format that is compatible with Unity.

[0064] However, when such an obj file is imported into Unity Simulation, a coordinate axis alignment problem occurs and readjustment is necessary. In the present invention, in order to rotate individual teeth, an image processing process is further performed in which the center point of the obj file format is set to the volume center point, and the coordinate values ​​of individual teeth are called from Rhino and placed on the tooth shape.

[0065] After the image processing process for arranging the tooth shape is performed, the first program, i.e., Unity, performs a process of saving the work data in the form of a script file.

[0066] Here, the working data is understood to be a series of treatment plan data related to the rearrangement of teeth, such as changing the arrangement of individual teeth that require orthodontic treatment, prepared by dental professionals, and such script files are also used as basic data in the application field for automating orthodontic treatment plans.

[0067] After step S40 is performed, an anomaly classification step (S50) is performed in which the relative arrangement state of the tooth objects included in the tooth arrangement data generated in step S40 is applied to an automatic diagnosis algorithm to automatically classify multiple anomalies including at least the shape of the malocclusion.

[0068] That is, in step S50, the tooth axis is derived based on the mesh information for each tooth for the tooth image obtained by executing steps S10 to S40, and the tooth image is processed to generate tooth data for each tooth, i.e., data including the feature data and the derived tooth axis.

[0069] On the other hand, for the axis of the tooth, a process is performed to derive the axis as shown in FIG.

[0070] For example, in the image processing step, for learning of individual teeth, the mesh information (T) for the entire tooth is processed as described above, and the volume center (VC) and the area center (AC) are automatically derived from the 3D data for each tooth, i.e., the mesh information (ET). The volume center (VC) is understood to mean the center point of the volume of each tooth data, and the area center (AC) is understood to mean, for example, the center point of the area of ​​the part with the widest area of ​​the tooth or the center point of the surface corresponding to the uppermost surface.

[0071] Here, the line passing through the volume center (VC) and the area center (AC) is primarily selected as the axis of the corresponding tooth. Meanwhile, according to the tooth image processing algorithm, an error correction function for the tooth data, e.g., the tooth axis, is derived through statistical analysis of the tooth image processing results. The derived numerical correction value is used to correct the passing line, and the corrected line is set as the tooth axis (A).

[0072] According to this, by using machine learning to analyze 3D tooth data, i.e., the pattern / movie-like features of the model and statistical analysis, an axis parallel to the tooth root is automatically derived and used as tooth data, and this is also used in simulations to establish treatment plans.

[0073] Meanwhile, as described above, mesh information of individual teeth can be converted into an obj format that can be used in simulations such as Unity. The tooth data is combined with this obj format. In this case, if an automation program for the Rhino 3D modeling tool is implemented using a program such as Python, all work processes using the tool can be automated with a script, making it possible to support automation services.

[0074] That is, by applying tooth coordinates and contact point and tooth axis data to each tooth image through this format conversion, data capable of representing the subject's tooth arrangement in three dimensions is generated as the tooth data and used for subsequent processing. This image format conversion is performed in the correction data generation step (S50) described below.

[0075] In addition, in step S50, the relative arrangement state of the tooth objects included in the tooth data generated in step S40 is applied to an automatic diagnosis algorithm to automatically classify multiple positional abnormalities including at least the shape of the malocclusion.

[0076] The relative arrangement state of tooth objects refers to the relative arrangement state of the teeth in the vertical and horizontal directions, which can be derived by comparing tooth data for adjacent teeth based on the tooth data, i.e., information on the coordinates, contact points, and axes of each tooth.

[0077] The automatic diagnostic algorithm is learned by the machine learning method, and the specific comparison process will be described later. That is, the learning of the automatic diagnostic algorithm can be understood to mean that the comparison standard values ​​to be used in the corresponding comparison process when automatically determining tooth position abnormalities through the comparison process described later are learned, or that the specific comparison process itself is learned.

[0078] An example of this is shown in detail in Figure 5ABCDEFG. First, A in Figure 5 indicates an abnormal classification corresponding to crowding. Crowding refers to a case where the mesiodistal contact points of adjacent teeth overlap, and is determined based on whether the teeth are arranged in an overlapping manner, i.e., whether the interdental contact points do not meet each other but cross each other.

[0079] If the result of the judgment is No, if the contact points meet and the arrangement is normal, it is judged as a normal arrangement, otherwise it is classified as a special situation, and the corresponding data is provided to medical staff and a special judgment instruction is requested. The result of the special judgment instruction is used for learning the automatic diagnosis algorithm. On the other hand, if there is no overlap and there is a gap, it proceeds to the gap evaluation described below, and if there is no overlap, it is also classified as a special situation.

[0080] On the other hand, if the result of the determination is Yes, the relative degree of crowding is determined based on the overlap degree. That is, if the overlap degree is -2 mm or less, the degree of crowding is determined to be 1 degree, if it is -2 to -4 mm, the degree of crowding is determined to be 2 degrees, and if it is -4 mm or more, the degree of crowding is determined to be 3 degrees.

[0081] Meanwhile, B in Figure 5 indicates an abnormality classification corresponding to the spacing. Spacing refers to the case where the mesiodistal contact points of adjacent teeth are separated from each other, leaving a space between the teeth. First, it is determined whether the teeth are spaced apart from each other, i.e., whether the interdental contact points are not overlapping and there is a space between them.

[0082] If the result of the judgment is No, if the contact points meet each other and the arrangement is normal, if there is no space and they overlap, the crowding evaluation is performed. On the other hand, if the contact points meet each other but the arrangement is not normal, or if there is no space and they do not overlap, it is classified as a special situation as described above.

[0083] On the other hand, if the answer is Yes, the relative gap size is determined based on the width of the space between them. For example, if the width of the space is 2 mm or less, it is determined to be a gap of 1 degree, if it is 2 to 4 mm, it is determined to be a gap of 2 degrees, and if it is 4 mm or more, it is determined to be a gap of 3 degrees.

[0084] C in Figure 5 indicates an abnormality classification corresponding to rotation. Rotation refers to the case where the mesiodistal contact point of the tooth is determined to be deviated from the normal alignment line, causing the tooth to rotate. In this case, the corresponding alignment line is used as a reference to determine whether the tooth has rotated.

[0085] If the result is No, and the contacts are aligned normally, it is judged as normal alignment, and if it is determined that there is rotation and a combination of gaps and crowding is present, it will be classified as a combination by evaluating both crowding and gaps. If the result is No, or the alignment is not normal, or there is no combination, it will be classified as a special case and special judgment instructions will be requested.

[0086] If the answer is yes, the degree of rotation is determined based on the relative rotation of the teeth. For example, if it is 4 degrees or less, it is determined as 1 degree rotation, if it is 4 to 8 degrees, it is determined as 2 degrees rotation, and if it is 8 degrees or more, it is determined as 3 degrees rotation.

[0087] D in Figure 5 indicates an abnormal classification corresponding to the vertical relationship (openbite & deepbite). The vertical relationship is an evaluation of the degree of overlap between upper and lower teeth, and all upper teeth must cover the lower teeth. If this relationship is not met, it is classified as an abnormal classification (D) corresponding to the vertical relationship, and a first value is preset as a normal value to match the characteristics of each tooth. The first value is also set by learning using the sample data.

[0088] In the classification process, it is determined whether the value matches the normal value or falls within the normal value range according to the characteristics of each tooth. If the result of the determination is Yes, the vertical overlap is normal compared to the reference value, and it is determined to be in a normal state. If not, it is classified as a special case as described above, and a special determination instruction is requested.

[0089] If the result of the judgment is No, it is judged whether the vertical coverage is insufficient compared to the standard value, and if it is insufficient, it is judged depending on the degree of insufficiency, for example, if it is 0 mm or less, it is judged as insufficiency level 1, if it is 0 to -3 mm, it is judged as insufficiency level 2, and if it is -3 mm or more, it is judged as insufficiency level 3. If the vertical coverage is not insufficient but is excessive, it is judged depending on the degree of excess, for example, if it is 2 mm or less than the standard value, it is judged as excessive level 1, if it is 2 mm to 4 mm, it is judged as excessive level 2, and if it is 4 mm or more, it is judged as excessive level 3.

[0090] E in Figure 5 indicates an abnormal classification corresponding to mesiodistal tooth axial inclination. Mesiodistal tooth axial inclination refers to a classification in which the slope of the tooth is excessively inclined mesial or distal on the mesiodistal surface of each tooth.

[0091] In this process, it is first determined whether the mesiodistal axial inclination gradient of each tooth matches a given second predetermined value, which is also set for each tooth by algorithm learning using the sample data.

[0092] If the result is Yes and the tooth axial inclination is within 2 degrees or normal, it is judged as normal; if not, it is classified as a special case and special judgment instructions are requested.

[0093] On the other hand, if the result of the judgment is No, it is judged whether the excessive tilt direction is mesial or not. If it is not mesial and the direction is not distal, it is classified as a special situation and special judgment instructions are requested. On the other hand, if it is mesial, if the degree of inclination is 4 degrees or less, it is judged as 1 degree mesial, if it is 4 degrees to 8 degrees, it is judged as 2 degrees mesial, and if it is 8 degrees or more, it is judged as 3 degrees mesial.

[0094] In the case of centrifugation, if the gradient is -4 degrees or less, the centrifugation degree is judged to be 1 degree, if it is between -4 degrees and -8 degrees, the centrifugation degree is judged to be 2 degrees, and if it is -8 degrees or more, the centrifugation degree is judged to be 3 degrees.

[0095] In Figure 5, "F" indicates an abnormality classification corresponding to the buccal-lingual tooth inclination. The buccal-lingual tooth inclination is a classification that determines whether the degree of inclination of the buccal-lingual surface of each tooth is excessive buccal inclination or excessive lingual inclination. First, in this process, it is determined whether the buccal-lingual tooth inclination gradient of each tooth matches a given third value. This third value is set to a value different from the mesiodistal tooth inclination, and can be set for each tooth and by algorithm learning using sample data.

[0096] If the answer is yes, and the tooth axial inclination is within 2 degrees or normal, it is judged as normal; otherwise, it is classified as a special case and special judgment instructions are requested.

[0097] On the other hand, if the judgment result is No, it is judged whether the excessive inclination direction is buccal, and if it is not buccal and the direction is not lingual, it is classified as a special situation and a special judgment instruction is requested. On the other hand, if it is buccal, if the inclination degree is 4 degrees or less, it is judged as 1 degree buccal, if it is 4 degrees to 8 degrees, it is judged as 2 degrees buccal, and if it is 8 degrees or more, it is judged as 3 degrees buccal.

[0098] In the case of the lingual side, depending on the degree of inclination, if it is -4 degrees or less, the lingual degree is judged to be 1 degree, if it is -4 degrees to -8 degrees, the lingual degree is judged to be 2 degrees, and if it is -8 degrees or more, the lingual degree is judged to be 3 degrees.

[0099] G in Figure 5 indicates an abnormality classification corresponding to occlusion. Occlusion is a classification determined by evaluating the occlusion state between the upper and lower teeth. It is determined based on whether the occlusion state between the teeth in a 1:2 relationship, where the teeth are biting together like sawtooth, is normal or not, and whether the cusps of the upper teeth are positioned on the cusps of the lower teeth or not.

[0100] If it is in an anterior (mesial) position, it is marked with "-", and if it is in a posterior (distal) position, it is marked with "+". First, in this process, the upper and lower teeth in the molar area must be arranged like sawtooth on the anterior-posterior (mesio-distal) plane, and it is determined whether the cusp of the upper tooth is sandwiched between two of the lower teeth.

[0101] If the result of the judgment is "Yes" and the fit of the upper and lower molars is normal, it is judged as a normal state; if not, it is classified as a special situation and a special judgment instruction is requested.

[0102] If the result of the judgment is No, it is judged whether the maxillary teeth are positioned forward, and if the maxillary teeth are positioned forward, the degree of mandibular backward position is judged as 1 degree if the degree is 2mm or less, 2 degrees if the degree is 2mm to 4mm, and 3 degrees if the degree is 4mm or more. On the other hand, if the maxillary teeth are positioned backward, the degree of mandibular forward position is judged as 1 degree if the degree is -2mm or less, 2 degrees if the degree is -2mm to -4mm, and 3 degrees if the degree is -4mm or more.

[0103] In this way, the basis for determining malocclusion is classified into seven categories, and the reference values ​​for each category are learned through sample data learning and real case learning, or the algorithm is learned through special judgment instructions from medical staff in special situations.This enables very accurate judgment based on the basis for determining malocclusion, and has the effect of enabling efficient and clear treatment solutions to be established in subsequent diagnoses.

[0104] In addition, in step S50, the degree of detail can be determined based on the magnitude of the numerical value that serves as the basis for determining each malocclusion, and can be set together with the malocclusion classification information, but the present invention is not limited to this.

[0105] After step S50 is performed, an orthodontic data generation step (S60) is performed in which each tooth image, each tooth data, each tooth identification information, and the classification results for each tooth positional abnormality are combined using the classification results of step S50 and the tooth data generated by executing step S40 to generate pre-treatment oral cavity condition information of the subject.

[0106] Here, in step S60, the pre-treatment oral cavity condition information of the subject includes, for example, tooth images as independent mesh information of each tooth, tooth data including feature data and axes, tooth identification information, and classification results for positional abnormalities classified by tooth. The corresponding data is stored, for example, in database 20, and is used for automatic treatment simulation or treatment plan establishment using AI.

[0107] Meanwhile, in order to use the corresponding pre-treatment oral condition information in the automatic treatment simulation or treatment plan establishment, orthodontic movement for each tooth must be possible, so as described above, it can be converted into an obj format that can be used in simulations such as Unity, etc. The tooth data is combined into this obj format.

[0108] That is, by converting the data into the above format by executing step S60, all processes using the tool can be automated by scripts by implementing a rhino 3D modeling tool automation program using a program such as Python, and automation services can be supported. By converting the format in this way, tooth coordinates, contact points, and tooth axis data are applied to each tooth image, and data that can represent the subject's tooth arrangement in 3D is generated from the tooth data and used for subsequent processing.

[0109] In another embodiment, after performing step S60 of FIG. 1, a process for recommending an orthodontic treatment plan may further include a pre-treatment oral condition information loading step (not shown) for loading pre-treatment oral condition information generated as a result of performing step S60, a predicted tooth arrangement model deriving step (not shown) for deriving one or more target predicted tooth arrangement models by considering the predicted direction and amount of movement of individual teeth included in the subject's pre-treatment oral condition information using an algorithm with previously learned learning data for tooth arrangement, an orthodontic treatment plan presenting step (not shown) for selecting an orthodontic solution for implementing the predicted tooth arrangement model and then presenting a step-by-step orthodontic treatment plan, and a feedback collecting step (not shown) for selecting one or more professionals who are matched to the subject's pre-treatment oral condition information from a professional talent pool in which a large number of professionals involved in the field of orthodontic treatment are registered, and collecting feedback on the presented orthodontic treatment plan.

[0110] Here, in the oral cavity condition information loading step, the pre-treatment oral cavity condition information is embodied as a 3D image as shown in FIG. 6AB.

[0111] Here, (a) of Figure 6 can be understood as an example of a simulation in which coordinate values ​​of individual teeth are not transmitted from Rhino, and (b) of Figure 6 can be understood as an example in which coordinate values ​​of individual teeth are transmitted from Rhino and the position, rotation angle, and size of each tooth are defined, corresponding to the subject's current dental condition, i.e., pre-treatment oral condition information is embodied in a 3D image.

[0112] In other words, in the present invention, by executing the oral condition information loading step, a 3D image that more clearly embodies the subject's current dental condition can be obtained, thereby increasing the precision of designing the orthodontic treatment plan described below.

[0113] Meanwhile, as a preferred embodiment of the present invention, in the execution of the oral condition information loading step, the present invention further collects image data including at least one of X-ray images, intraoral images, and facial images for the subject's pre-treatment oral condition information as reference data, and uses this as a condition variable for the candidate algorithm described below.

[0114] More specifically, the X-ray image is understood to be an image including a cephalogram, which is an X-ray image of the side of the face containing the most important information in orthodontic treatment, and a panoramic image, which is an X-ray image of a wide range of scenes taken by moving the camera from one side to the other or from top to bottom. The intraoral image is an image obtained by simply taking a 2D image of the inside of the oral cavity using a camera, etc., and the facial image is understood to be an image that shows the facial features of the subject depending on the oral structure and dental condition.

[0115] Here, in the present invention, the horizontal and vertical relationship of the upper and lower jaws can be measured from the cephalometric images collected as reference data, and the degree of maxillary protrusion and lingual protrusion can be measured. Also, missing teeth can be identified from the panoramic images, and the health of the alveolar bone can be quantified and determined as a numerical value, and excessive tooth inclination, the presence or absence of wisdom teeth / cavities, and the number of implants can be measured.

[0116] In addition, the intraoral photographs can be used to measure the number of cavities, fillings, and crowns, and the facial photographs can be used to perform horizontal, vertical, and transverse evaluations.

[0117] In other words, the present invention has the effect of converting non-standard diagnostic data into a database in the form of standard data through the collection of the reference data and the measurement values ​​of the collected reference data, thereby enabling the classification and definition of standard diagnostic data for orthodontic treatment plans to be designed.

[0118] Next, the predicted tooth arrangement model referred to in the step of deriving the predicted tooth arrangement model is derived as a model that is closest to a standard model having an ideal tooth arrangement in the field of orthodontic treatment. Here, the ideal tooth arrangement is understood to mean a tooth arrangement in which the contact points of a tooth and the contact points of an adjacent tooth are in abutting state without overlapping, the mesiodistal contact points of adjacent teeth are in abutting state, the tooth rotation angle on the occlusal surface is within a normal range, in terms of the degree of occlusion of the upper and lower teeth, all maxillary teeth cover the mandibular teeth, the tooth inclination on the mesiodistal surface of each tooth is within a normal range, the tooth inclination on the buccal and lingual surface of each tooth is normal, and the fit of the upper and lower molars is tilted to one side or is not a 1:2 relationship occlusion.

[0119] That is, the algorithm can machine-learn learning data for ideal tooth arrangement and derive a predicted tooth arrangement model that provides the most ideal tooth arrangement for the subject's current dental condition.

[0120] Meanwhile, the predicted direction and amount of tooth movement are predicted based on clinical data on orthodontic treatment, and it is understood that the direction and amount of tooth movement can be predicted using a weak tool and the RF (Random Forest) algorithm, which is one of the meta-learning configuration algorithms.

[0121] Here, meta-learning is a mechanism for deriving the best model from among a number of learning models, and 400 in Figure 9 shows a conceptual diagram of meta-learning. The meta-learner in 400 in Figure 9 uses an algorithm including at least one of the RF, ANN (Artificial Neural Network), RF (Random Forest), SVM (Support Vector Machine), and EDN (Evolving Deep Network).

[0122] Also, referring to Figure 7AB, (a) of Figure 7 shows an example of a 3D image corresponding to the subject's pre-treatment oral condition information, and (b) of Figure 7 shows an example of a predicted tooth arrangement model derived to be closest to the ideal tooth arrangement based on the subject's current dental condition.

[0123] Here, in FIG. 7(b), it can be seen that the subject's tooth arrangement has become more uniform compared to FIG. 7(a) due to the algorithm that learned the ideal tooth arrangement.

[0124] On the other hand, in general, in order to present an orthodontic treatment plan, it is necessary to search for a solution that simultaneously satisfies conflicting factors such as the duration of orthodontic treatment, the orthodontic force, and the difficulty of orthodontic treatment.

[0125] In this invention, in the execution of the orthodontic treatment plan presentation step, a Pareto optimal solution calculation method is used to set the predicted value of elements including at least one (preferably all) of the orthodontic treatment period, orthodontic force, and difficulty of orthodontic treatment based on the current dental condition of the subject as the objective function, and a Pareto optimal solution is searched for that simultaneously satisfies these objective functions and minimizes the value of the objective function, thereby presenting multiple orthodontic treatment plans.

[0126] [Table 1] Dominance relationships and Pareto optimal solutions

[0127] As an example, referring to Table 1 above, in the case of a minimization problem, a vector having n decision variables

number

number

number

number

[0128] Also, Pareto optimality is defined as the vector F(x), where D is the decision space of all possible x evaluated by the vector F(x).

number

number

number

number

number

number

[0129] By the above definition, individual F can be said to dominate the region to which individual G belongs, and again, F is dominated by individuals B and C.

[0130] As a result, individuals A, B, C, and D are in an equal relationship that cannot be compared with each other, and are not dominated by any individual, resulting in a Pareto optimal solution.

[0131] In addition, in order to find a Pareto-optimal solution that simultaneously optimizes multiple objective functions, a multiple optimization algorithm is used to generate an arbitrary N number of initial populations and assign them a Pareto ranking in a non-dominated order.

[0132] That is, a rank of 1 is assigned to a candidate solution that is not dominated by any other individual in the entire population in the multi-criteria evaluation, and then the rank-1 candidate solution is removed from the entire population. Then, a rank-2 non-dominated individual is again selected from the remaining population and assigned a rank.

[0133] By repeating this process, all candidate solutions will have been assigned a rank, and in the next step, genetic operations consisting of natural selection, crossover, and mutation will be applied in order to generate new candidate solutions.

[0134] Here, in the natural selection step, all candidate solutions of rank 1 are included, and for the remaining candidate solutions, two candidate solutions are arbitrarily selected, of which the candidate solution with the higher rank is included. In the case of candidate solutions of the same rank, the solution with the lowest density of individuals in the evaluation space of the multiple objective function can be selected as the priority, and the non-dominated solution in the final step can be provided as the proposed orthodontic treatment plan.

[0135] In the orthodontic treatment plan presentation step, when multiple orthodontic treatment plans are derived using the Pareto optimal solution calculation method, a candidate algorithm including at least one of ANN (Artificial Neural Network), RF (Random Forest), SVM (Support Vector Machine), and EDN (Evolving Deep Network) is used to calculate the optimal orthodontic treatment plan for the subject's current dental condition from among the multiple orthodontic treatment plans.

[0136] Here, the ANN is an algorithm that mimics the neural network structure of the brain and is well known as deep learning. When a large number of instances at the big data level are secured, it has the advantage of being able to induce high performance by applying techniques such as increasing the number of layers, structural changes, abstraction, and drop-out.

[0137] In addition, RF has the advantage of inducing collective intelligence by merging the prediction values ​​of multiple weak classifiers to derive a single result, thereby achieving higher performance than a single classifier. Furthermore, RF can generate a model very quickly in the case of a weak classifier consisting of a decision tree.

[0138] Furthermore, the SVM is a method of setting the data class region using a support vector and deriving the boundary line. In other words, when a set of data belonging to one of two categories is given, the SVM has the advantage of being able to generate a non-probabilistic binomial linear classification model that determines which category new data belongs to based on the given data set.

[0139] In addition, EDN is a method in which evolutionary computation is applied to a deep network learning algorithm, and has the advantage of being able to achieve fast learning and processing speed and improved accuracy based on a large number of weak classifiers.

[0140] That is, in the present invention, the candidate algorithm is used to apply evolutionary combination selection of a large number of attribute data to maximize diversity, and the performance of each weak classifier is improved through a deep network, thereby inducing the expression of collective intelligence, thereby enabling the establishment of an advanced orthodontic treatment plan.

[0141] Meanwhile, the optimal orthodontic treatment plan derived in this way for the subject's current dental condition is embodied in an interface such as 500 in Figure 10, and an orthodontic treatment plan that fits the subject can be provided in stages.

[0142] Here, the interface processes the time-series tooth movement path to realize a predicted tooth arrangement model based on the calculated optimal orthodontic treatment plan and the subject's pre-treatment oral condition information into a time-series model and provides it as shown in 510 in Figure 10. In the present invention, by performing such functions, the predicted results of the subject's orthodontic treatment can be more intuitively considered.

[0143] In another preferred embodiment, the feedback collection step is performed after the orthodontic treatment plan presentation step is performed.

[0144] Here, the term "experts" is limited to those who have acquired orthodontic and specialist qualifications in order to improve the quality of orthodontic treatment, and it is desirable to select experts whose clinical experience corresponding to the subject's pre-treatment oral condition information is judged to be above a certain critical standard. The criteria for judging clinical experience are that experts upload cases to the talent pool to reveal their own experience, write down their findings on the case, and then have the validity of the case evaluated by other experts or external experts designated as case review committee members, thereby managing the clinical experience of experts.

[0145] In addition, in the execution of the feedback collection step, the number of specialists who are matched with the subject's pre-treatment oral condition information is set to one or more, or five or less, to ensure smooth collaboration between the specialists.

[0146] In addition, in the feedback collection step, once a professional who matches the subject's pre-treatment oral condition information is determined, feedback on the orthodontic treatment plan presented in the orthodontic treatment plan presentation step is collected, and in the present invention, the feedback collected from one or more professional is labeled so that supervised learning can be performed on the algorithm.

[0147] In other words, in the present invention, when an orthodontic treatment plan is derived using an algorithm based on machine learning, specialized personnel can learn from the presented results of problems that have not yet been found and the solutions that are treatment plans for those problems, thereby improving the sophistication of the orthodontic treatment plan that is presented subsequently.An instance configuration method for applying such machine learning can be seen at 300 in Figure 8.

[0148] Here, the Before Array in Figure 8 is understood to be data representing the subject's pre-treatment oral condition information, i.e., the current dental condition, the Target Array is data corresponding to the subject's predicted tooth arrangement model, and the Solution Label is data corresponding to the final orthodontic treatment plan obtained by labeling the feedback collected from professional personnel.

[0149] Meanwhile, in another embodiment of the present invention, after the feedback collection step is performed and a final orthodontic treatment plan for the subject is determined, the computing device may perform a function of creating and processing a treatment plan for the final orthodontic treatment plan in a predetermined file format (including at least one of a PDF and a video file format) and transmitting it to a business entity terminal including at least one of the dental clinic visited by the subject and the orthodontic device manufacturer.

[0150] That is, the above embodiment can be understood as a process of documenting an orthodontic treatment plan such as 300 and 310 in FIG. 10, and transmitting such a treatment plan to the terminal of the dental clinic visited by the patient allows the orthodontic treatment based on the treatment plan to be transferred at the dental clinic visited by the patient, thereby reducing the time required for orthodontic treatment diagnosis and orthodontic treatment planning, improving diagnostic efficiency, and reducing quality deviations due to differences in orthodontic treatment results at each dental clinic.

[0151] On the other hand, it is desirable that the orthodontic treatment plan be sent to the orthodontic device manufacturer's terminal after confirmation by the visiting dentist.More preferably, the orthodontic device manufacturer is requested to manufacture orthodontic devices according to the time period, taking into consideration the number of visits to the hospital or the timing of the treatment plan in the subject's treatment plan, so that the orthodontic devices can be manufactured in a timely manner, thereby achieving the goal of efficient orthodontic treatment.

[0152] Next, Figure 11 shows a configuration diagram of an apparatus for separating tooth objects from 3D oral cavity scan data, automatically detecting tooth positional abnormalities, and recommending an orthodontic treatment plan according to one embodiment of the present invention. In the following explanation, explanations that overlap with those of Figures 1 to 7AB will be omitted.

[0153] As shown in Figure 11, in the present invention, the main component of the device 10 includes an initial contact point search unit 11 that functions to sequentially increase the size of a circle at the simple coordinate center of the 3D oral scan data 1 corresponding to the subject's pre-treatment oral condition information, and search for the first point where the circle and an object included in the 3D oral scan data 1 intersect.

[0154] Here, it is understood that the initial contact point search unit 11 can ultimately execute any of the functions performed by step S10 in Figure 1, and in the present invention, the execution of the functions of the initial contact point search unit 11 determines the position of the tooth object in the 3D oral cavity scan data 1.

[0155] In addition, in the present invention, the main components of the device 10 include a sphere placement unit 12 that functions to distribute one or more spheres at a predetermined interval within the tooth at the initial contact point detected by the initial contact point detection unit 11, then reverse the vector direction of the mesh, move the spheres toward the periphery in the reference coordinates at a random speed, and replicate the spheres at a predetermined period so that the spheres are uniformly distributed within the tooth.

[0156] In other words, the sphere placement unit 12 can be understood to be capable of performing all of the functions performed by step S20 in Figure 1, and in the present invention, a method is used in which a large number of small spheres having a size range of 1 to 5 units are distributed within the tooth in the sphere placement unit 12, and feature points of individual teeth are derived, thereby achieving the effect of enabling extraction of physical characteristics of teeth in a method different from the conventional method of extracting tooth feature points based on the outer shape of the tooth.

[0157] In addition, in the present invention, the main components of the device 10 include a feature data extraction unit 13 that grows the size of the spheres replicated by the sphere placement unit 12 to a predetermined size, determines a representative sphere corresponding to each tooth, and extracts tooth feature data based on the coordinate points and contact points of the determined representative sphere.

[0158] In other words, the feature data extraction unit 13 can be understood to perform all of the functions performed by step S30 in Figure 1. In the present invention, by performing the functions of the feature data extraction unit 13, it is possible to extract tooth feature data based on coordinate system data detection to which the physical characteristics of the tooth are applied, unlike the conventional method in which tooth feature data is extracted based on the outer shape of the tooth, thereby providing an advantageous effect of proposing a new approach to tooth feature data extraction.

[0159] Here, it is preferable that the tooth feature data extracted by the feature data extracting unit 13 is stored and managed in a dedicated database 30, but the present invention is not limited to this.

[0160] In addition, the present invention includes, as a main component of the device 10, an image processing unit 14 that functions to obtain mesh information of individual teeth using feature data extracted for each tooth and axis data for each tooth, thereby embodying the subject's tooth arrangement in an individually separated state as a 3D image.

[0161] In other words, it is understood that the image processing unit 14 is capable of executing all of the functions performed by step S40 in Figure 1, and in the present invention, by executing the functions of the image processing unit 14, it is possible to design possible movement paths for individual teeth and to have the image processing technology that visualizes the possible movement paths on the external terminal 20 be algorithmized.

[0162] In addition, in the present invention, the main component of the device 10 includes a positional abnormality classification unit 15 that applies the relative arrangement state of tooth objects included in the tooth arrangement data generated by the image processing unit to an automatic diagnosis algorithm to automatically classify multiple positional abnormalities including at least the shape of malocclusion.

[0163] Here, the positional abnormality classification unit is understood to perform all of the functions performed by step S50 in Figure 1. In the present invention, by performing the functions of such positional abnormality classification unit 15, it is possible to very accurately determine whether or not positional abnormality exists based on the subject's dental data as a basis for determining malocclusion, which has the effect of enabling efficient and clear treatment solutions to be established in subsequent diagnoses.

[0164] In addition, the present invention includes, as a main component of the device 10, an orthodontic data generation unit 16 that functions to combine each tooth image, each tooth data, each tooth identification information, and the classification result for each tooth positional abnormality based on the function execution result of the positional abnormality classification unit, and generate pre-treatment oral condition information of the subject.

[0165] As a result, it is understood that the correction data generation unit is capable of executing all of the functions performed by step S60 in Figure 1, and in the present invention, the execution of the functions of such correction data generation unit 16 has the effect of facilitating automatic treatment simulation or establishment of a treatment plan.

[0166] Furthermore, as shown in Figure 11, in the present invention, the main components of the device 10 further include a pre-treatment oral cavity condition information loading unit 17 that loads pre-treatment oral cavity condition information generated as a result of executing the function of the orthodontic data generation unit 16.

[0167] Here, the pre-treatment oral cavity condition information loading unit performs a function of loading the pre-treatment oral cavity condition information of the subject in a first program (for example, a Unity simulation program) which is a simulation program to which a physics engine is applied.

[0168] In the present invention, the function of the pre-treatment oral cavity condition information loading unit 17 is executed to embody the tooth arrangement structure corresponding to the subject's pre-treatment oral cavity condition information in a 3D image. In particular, unlike the conventional method in which individual teeth were separated manually from 3D oral cavity scan data by professionals, the present invention embody a 3D image in which individual teeth are separated with high accuracy from the subject's 3D oral cavity scan data using feature point data and axis data of each tooth, thereby increasing the convenience and efficiency of establishing an orthodontic treatment plan.

[0169] As shown in FIG. 11, the present invention also includes, as a main component of the device 10, a predicted tooth arrangement model derivation unit 18 that derives an optimal predicted tooth arrangement model from a database 31 in which one or more target predicted tooth arrangement models are stored, by taking into account the predicted direction and amount of movement of each tooth included in the subject's pre-treatment oral condition information, using an algorithm that has already learned learning data for tooth arrangement.

[0170] In other words, in the present invention, by executing the functions of the predicted tooth arrangement model derivation unit 18, it is possible to derive a predicted tooth arrangement model that is closest to the most ideal tooth model, taking into account the subject's pre-treatment oral condition information.

[0171] In addition, as shown in FIG. 11, the present invention includes, as a main component of the device 10, an orthodontic treatment plan presentation unit 19 that selects an orthodontic solution for implementing a predicted tooth arrangement model and then presents a step-by-step orthodontic treatment plan.

[0172] In other words, in the present invention, the orthodontic treatment plan presentation unit 19 performs its functions to search for an optimal solution that simultaneously satisfies conflicting elements such as the duration of orthodontic treatment, the orthodontic force, and the difficulty of orthodontic treatment, and presents multiple orthodontic treatment plans. The optimal orthodontic treatment plan for the subject's pre-treatment oral condition information is calculated from the multiple orthodontic treatment plans using a candidate algorithm including at least one of ANN (Artificial Neural Network), RF (Random Forest), SVM (Support Vector Machine), and EDN (Evolving Deep Network), thereby enabling the establishment of an advanced orthodontic treatment plan.

[0173] In addition, as shown in FIG. 11, in the present invention, the main components of the device 10 further include a feedback collection unit 20 that selects one or more experts 40 that match the subject's pre-treatment oral condition information from a professional talent pool in which a large number of experts involved in the field of orthodontic treatment are registered, and collects feedback on the presented orthodontic treatment plan.

[0174] That is, in the present invention, by executing the functions of the feedback collection unit 20, it is possible to apply objective and quantitative artificial intelligence based on machine learning and big data for diagnosis, and to present an advanced treatment plan based on a large amount of clinical experience when establishing a treatment plan.

[0175] Although the present invention has been described above with reference to the limited embodiments and drawings, those skilled in the art will appreciate that various modifications and variations can be made from the above description.

[0176] Next, FIG. 12 shows an example of the internal configuration of a computing device according to an embodiment of the present invention, and in the following description, any description that overlaps with the description of FIGS. 1 to 11 will be omitted.

[0177] 12, the computing device 10000 includes at least one processor 11100, a memory 11200, a peripheral device interface 11300, an input / output subsystem 11400, a power circuit 11500, and a communication circuit 11600. Here, the computing device 10000 corresponds to a user terminal (A) connected to a haptic interface device or the computing device (B).

[0178] The memory 11200 may include, for example, high-speed random access memory, a magnetic disk, an SRAM, a DRAM, a ROM, a flash memory, or a non-volatile memory. The memory 11200 may include software modules, command sets, or various other data required for the operation of the computing device 10000.

[0179] Here, access to the memory 11200 from other components such as the processor 11100 and the peripheral device interface 11300 is controlled by the processor 11100.

[0180] The peripheral interface 11300 couples input and / or output peripherals of the computing device 10000 to the processor 11100 and memory 11200. The processor 11100 executes software modules or sets of commands stored in the memory 11200 to perform various functions and process data for the computing device 10000.

[0181] The I / O subsystem 11400 couples various I / O peripherals to the peripheral interface 11300. For example, the I / O subsystem 11400 includes controllers for coupling peripherals such as a monitor, keyboard, mouse, printer, and, if desired, a touch screen or sensor to the peripheral interface 11300. In another aspect, I / O peripherals can also be coupled to the peripheral interface 11300 without going through the I / O subsystem 11400.

[0182] The power circuit 11500 can supply power to all or some of the components of the terminal device. For example, the power circuit 11500 can include a power management system, one or more power sources such as a battery or alternating current (AC), a charging system, a power failure detection circuit, a power converter or inverter, a power status indicator, or any other component for power generation, management, or distribution.

[0183] The communications circuitry 11600 allows for communication with other computing devices using at least one external port.

[0184] Alternatively, as mentioned above, if desired, the communications circuitry 11600 may include RF circuitry to enable communication with other computing devices by sending and receiving RF signals, also known as electromagnetic signals.

[0185] 12 is merely one example of computing device 10000, and computing device 11000 may omit some of the components in FIG. 12, include additional components in FIG. 12, or have a configuration or arrangement in which two or more components are combined. For example, a computing device for a communication terminal in a mobile environment may further include a touch screen, a sensor, etc. in addition to the components shown in FIG. 12, and communication circuit 1160 may include circuits for RF communication of various communication methods (Wi-Fi, 3G, LTE, Bluetooth (registered trademark), NFC, Zigbee (registered trademark), etc.). The components included in computing device 10000 may be embodied as hardware, software, or a combination of both hardware and software, including one or more signal processing or application-specific integrated circuits.

[0186] Methods according to embodiments of the present invention may be implemented in the form of program instructions executed by various computing devices and recorded on a computer-readable medium. In particular, programs according to embodiments of the present invention may be implemented as PC-based programs or applications dedicated to mobile terminals. Applications to which the present invention is applied are installed in user terminals through files provided by a file distribution system. For example, the file distribution system may include a file transmission unit (not shown) that transmits the files at the request of the user terminal.

[0187] The devices described above may be implemented using hardware components, software components, and / or a combination of hardware and software components. For example, the devices and components described in the embodiments may be implemented using one or more general-purpose or special-purpose computers, such as a processor, controller, arithmetic logic unit (ALU), digital signal processor, microcomputer, field programmable gate array (FPGA), programmable logic unit (PLU), microprocessor, or any other device that executes and responds to instructions. The processing device may execute an operating system (OS) and one or more software applications that run on the operating system.

[0188] A processing device may also access, store, manipulate, process, and generate data in response to the execution of software. For ease of understanding, a single processing device may be described. However, those skilled in the art will recognize that a processing device may include multiple processing elements and / or multiple types of processing elements. For example, a processing device may include multiple processors or a processor and a controller. Other processing configurations, such as parallel processors, are also possible.

[0189] Software includes computer programs, codes, instructions, or a combination of one or more of these, which can configure or independently or collectively instruct a processing device to operate as desired. The software and / or data can be permanently or temporarily embodied in any type of machine, component, physical device, virtual device, computer storage medium, or device to be analyzed by the processing device or to provide instructions or data to the processing device. The software can also be distributed across network-coupled computing devices and stored or executed in a distributed manner. The software and data can be stored on one or more computer-readable recording media.

[0190] Methods according to the embodiments may be embodied in the form of program instructions executed by various computer means and recorded on a computer-readable medium. The computer-readable medium may include, alone or in combination, program instructions, data files, data structures, and the like. The program instructions recorded on the medium may be those specially designed and constructed for the embodiments, or they may be those known and available to those skilled in the art of computer software. Computer-readable media include magnetic media such as hard disks, floppy disks, and magnetic tape, optical media such as CD-ROMs and DVDs, magneto-optical media such as floptical disks, and hardware devices specially configured to store and execute program instructions, such as ROM, RAM, flash memory, and the like.

[0191] Program instructions include not only machine code, such as produced by a compiler, but also high-level language code that can be executed by a computer using an interpreter, etc. The hardware devices may be configured to operate as one or more software modules to perform the operations of an embodiment, and vice versa.

[0192] Although the above embodiments have been described using only limited examples and drawings, those skilled in the art will appreciate that various modifications and variations may be made from the above description. For example, the described techniques may be performed in a different order than described, and / or the described system, structure, device, circuit, or other components may be combined or combined in a different manner than described, or may be substituted or replaced by other components or equivalents, while still achieving suitable results. Therefore, other implementations, other embodiments, and equivalents to the claims are also within the scope of the following claims.

Claims

1. 1. A method for separating dental objects from three-dimensional oral scan data, automatically identifying dental malpositions, and recommending an orthodontic treatment plan, the method being implemented in a computing device including one or more processors and a main memory storing instructions executable by the processors, the method comprising: an initial contact point search step of sequentially increasing the size of a circle at the center of the simple coordinates of the 3D oral cavity scan data corresponding to the subject's pre-treatment oral cavity condition information, and searching for an initial point where the circle and a tooth object included in the 3D oral cavity scan data contact each other; a sphere placement step in which one or more spheres are distributed within the tooth at a predetermined interval from the first contact point found, and then the vector direction of the mesh is reversed, and the spheres are moved at a random speed toward the periphery of the reference coordinate system while duplicating the spheres at a predetermined cycle, so that the spheres are uniformly placed within the tooth; a feature data extraction step of growing the replicated spheres in the sphere placement step to a predetermined size to determine a representative sphere corresponding to each tooth, and extracting feature data of the tooth based on the coordinates and contact points of the determined representative sphere; an image processing step of obtaining mesh information of individual teeth using the feature data extracted for each tooth and the tooth-specific axis data, thereby realizing the subject's tooth arrangement in an individually separated state as a 3D image and generating tooth arrangement data; a positional anomaly classification step of automatically classifying a plurality of positional anomalies including at least the shape of malocclusion by applying the relative positional anomalies of the tooth objects included in the tooth arrangement data generated in the image processing step, which are derived by comparing tooth data for the adjacent teeth based on the coordinates and the contact points of the representative spheres of each tooth object and the tooth-specific axis data, to an automatic diagnosis algorithm obtained by deep learning as a result of determining whether the relative positional anomalies correspond to at least one of crowding, spacing, rotation, perpendicular relationship (openbite & deepbite), mesiodistal inclination (tipping), buccolingual inclination (torque), and occlusion; and and a step of generating orthodontic data based on the result of the execution of the positional abnormality classification step by combining each tooth image, each tooth data, each tooth identification information, and the classification result for each tooth positional abnormality, and generating the orthodontic data based on the subject's pre-treatment oral condition information.

2. In the image processing step, The tooth-specific axis data is Extracted based on the volume center point and the area center point of the 3D oral cavity scan data, A method for separating tooth objects from 3D oral scan data, automatically detecting tooth positional abnormalities, and recommending an orthodontic treatment plan, as described in claim 1, characterized in that a numerical correction value derived by statistical analysis is applied to a line passing through the volume center point and the width center point to extract tooth-specific axial data acquired parallel to the tooth roots.

3. The determination of the representative sphere is The size of the replicated spheres is increased to a predetermined size, and the spheres that have fallen out of the tooth are dropped off, and then the spheres remaining in the tooth are reduced to a predetermined size so that a representative sphere corresponding to the center point of each tooth is determined; A method for separating tooth objects from three-dimensional oral scan data, automatically detecting tooth positional abnormalities, and recommending an orthodontic treatment plan, as described in claim 1, characterized in that the method is continued until a representative sphere corresponding to the number of teeth of the subject is obtained.

4. 2. The method for separating tooth objects from 3D oral scan data, automatically detecting tooth positional abnormalities, and recommending an orthodontic treatment plan according to claim 1, wherein the image processing step converts and exports the mesh information of the individual teeth into a predetermined file format compatible with a first program that handles motion data, and then saves the work data in the first program as a script file.

5. 2. The method for separating tooth objects from 3D oral scan data, automatically detecting tooth positional abnormalities, and recommending an orthodontic treatment plan according to claim 1, wherein the feature data extraction step extracts x, y, and z coordinate values ​​for position in 3D space, x, y, and z coordinate values ​​for rotation, and x, y, and z coordinate values ​​for size as feature data of a representative sphere corresponding to each individual tooth.

6. The method for separating tooth objects from 3D oral scan data, automatically detecting tooth positional abnormalities, and recommending an orthodontic treatment plan, as described in claim 1, characterized in that, in performing the sphere placement step, the size of the spheres dispersed within the tooth has a size range of 1 to 5 units, and 1 unit corresponds to 1 / 10,000 mm.

7. 2. The method of claim 1, wherein the positional abnormality classification step includes, based on the size and relative arrangement of the teeth, classifying the positional abnormality as crowding if interdental contact points do not meet but intersect, classifying the positional abnormality as void if there is a space between the interdental contact points, classifying the positional abnormality as rotated if the teeth are rotated relative to each other, classifying the positional abnormality as perpendicular if the perpendicular relationship of the teeth does not match a predetermined first value, classifying the positional abnormality as mesiodistal inclination if the gradient of the mesiodistal axial inclination of the teeth does not match a predetermined second value, classifying the positional abnormality as buccolingual inclination if the gradient of the buccolingual axial inclination of the teeth does not match a predetermined third value, and classifying the positional abnormality as interdigitation if the cusp of a maxillary tooth is not located between two mandibular teeth.

8. The method for separating tooth objects from three-dimensional oral scan data, automatically detecting tooth positional abnormalities, and recommending an orthodontic treatment plan, as described in claim 7, wherein the positional abnormality classification step determines the level of detail based on the size of the numerical value that serves as the basis for judgment for each malocclusion, and sets it together with the malocclusion classification information.

9. A method for separating tooth objects from three-dimensional oral cavity scan data, automatically detecting tooth positional abnormalities, and recommending an orthodontic treatment plan, as described in claim 1, characterized in that in the orthodontic data generation step, mesh information of tooth images derived from three-dimensional oral cavity scan data is converted into object format that can be simulated by a first program, and data is generated together with the pre-treatment oral cavity condition information.

10. After the correction data generation step is executed, a pre-treatment oral cavity condition information loading step of loading pre-treatment oral cavity condition information generated as a result of execution of the orthodontic data generating step; a predicted tooth arrangement model deriving step for deriving one or more targeted predicted tooth arrangement models by considering the predicted direction and amount of movement of each individual tooth included in the subject's pre-treatment oral condition information according to an algorithm with previously learned learning data for the tooth arrangement; an orthodontic treatment plan presenting step of selecting an orthodontic solution for implementing the predicted tooth arrangement model and then presenting a step-by-step orthodontic treatment plan; The method for separating tooth objects from three-dimensional oral scan data, automatically detecting tooth positional abnormalities, and recommending an orthodontic treatment plan according to claim 1, further comprising a feedback collection step of selecting one or more experts who match the subject's pre-treatment oral condition information from a talent pool of experts in the field of orthodontic treatment, and collecting feedback on the presented orthodontic treatment plan.

11. The orthodontic treatment plan presentation step includes: The objective function is a predicted value of an element including at least one of the orthodontic treatment period, orthodontic strength, and difficulty of orthodontic treatment according to the subject's current dental condition, 11. The method for separating tooth objects from three-dimensional oral cavity scan data, automatically detecting tooth positional abnormalities, and recommending an orthodontic treatment plan according to claim 10, wherein a Pareto optimal solution calculation method is used to search for Pareto optimal solutions that simultaneously satisfy the objective functions and minimize the values ​​of the objective functions, thereby presenting a plurality of orthodontic treatment plans; and when a plurality of orthodontic treatment plans are derived by the Pareto optimal solution calculation method, an optimal orthodontic treatment plan from the plurality of orthodontic treatment plans for the subject's pre-treatment oral condition information is calculated using a candidate algorithm including at least one of ANN (Artificial Neural Network), RF (Random Forest), SVM (Support Vector Machine), and EDN (Evolving Deep Network).

12. In the execution of the pre-treatment oral condition information loading step, Further collecting image data including at least one of an X-ray image, an intraoral image, and a facial image regarding the subject's pre-treatment oral condition information as reference data; The method for separating tooth objects from 3D oral scan data, automatically detecting tooth positional abnormalities, and recommending an orthodontic treatment plan, as described in claim 11, wherein the reference data is used only as a condition variable for the candidate algorithm.

13. The method for separating tooth objects from three-dimensional oral cavity scan data, automatically detecting tooth positional abnormalities, and recommending an orthodontic treatment plan, as described in claim 11, characterized in that the orthodontic treatment plan presenting step processes a time-series tooth movement path for realizing the predicted tooth arrangement model in the subject's pre-treatment oral condition information using a time-series model and provides it based on the calculated optimal orthodontic treatment plan.

14. In performing the feedback collection step, The selection of the specialists is performed by selecting specialists whose clinical history corresponding to the subject's pre-treatment oral condition information is determined to be equal to or greater than a predetermined critical standard; In the feedback collection step, The method for separating tooth objects from 3D oral scan data, automatically detecting tooth malpositions, and recommending an orthodontic treatment plan according to claim 10, further comprising labeling the feedback collected from one or more experts to enable supervised learning for the algorithm.

15. 1. An apparatus for separating dental objects from 3D oral scan data, automatically detecting tooth malpositions, and recommending an orthodontic treatment plan, the apparatus being implemented in a computing device including one or more processors and a main memory storing instructions executable by the processors, the apparatus comprising: an initial contact point search unit that sequentially increases the size of a circle at the center of a simple coordinate of the 3D oral cavity scan data corresponding to the subject's pre-treatment oral cavity condition information, and searches for an initial point where the circle and a tooth object included in the 3D oral cavity scan data contact each other; a sphere placement unit that distributes one or more spheres in the tooth at a predetermined interval from the first contact point found, reverses the vector direction of the mesh, moves the spheres in a circumferential direction at a random speed in the reference coordinate system, and replicates the spheres at a predetermined cycle, thereby distributing the spheres uniformly in the tooth; a feature data extraction unit that grows the size of the spheres replicated by the sphere placement unit to a predetermined size to determine a representative sphere corresponding to each tooth, and extracts feature data of the tooth based on the coordinates and contact points of the determined representative sphere; an image processing unit that obtains mesh information of individual teeth using the feature data extracted for each tooth and the tooth-specific axis data, thereby realizing the subject's tooth arrangement in an individually separated state as a 3D image and generating tooth arrangement data; a positional anomaly classification unit that automatically classifies a plurality of positional anomalies, including at least the shape of malocclusion, by applying the relative positional states of tooth objects included in the tooth arrangement data generated by the image processing unit, derived by comparing tooth data for adjacent teeth based on the coordinates and the contact points of the representative spheres of each tooth object and the tooth-specific axis data, to an automatic diagnosis algorithm obtained by deep learning that determines whether the relative positional state of tooth objects corresponds to at least one of positional anomalies of crowding, spacing, rotation, perpendicular relationship (openbite & deepbite), mesiodistal tooth inclination (tipping), buccolingual tooth inclination (torque), and occlusion; and an orthodontic data generation unit that generates orthodontic data based on the subject's pre-treatment oral condition information by combining each tooth image, each tooth data, each tooth identification information, and classification results for positional abnormalities for each tooth according to the function execution result of the positional abnormality classification unit.

16. A computer-readable recording medium, The computer-readable medium stores instructions that cause a computing device to perform the following steps, the steps comprising: an initial contact point search step of sequentially increasing the size of a circle at the center of the simple coordinates of the 3D oral cavity scan data corresponding to the subject's pre-treatment oral cavity condition information, and searching for an initial point where the circle and a tooth object included in the 3D oral cavity scan data contact each other; a sphere placement step in which one or more spheres are distributed within the tooth at a predetermined interval from the first contact point found, and then the vector direction of the mesh is reversed, and the spheres are moved at a random speed toward the periphery of the reference coordinate system while duplicating the spheres at a predetermined cycle, so that the spheres are uniformly placed within the tooth; a feature data extraction step of determining a representative sphere corresponding to each tooth by growing the replicated spheres in a predetermined size in the sphere placement step, and extracting feature data of the tooth based on the coordinates and contact points of the determined representative sphere; an image processing step of obtaining mesh information of individual teeth using the feature data extracted for each tooth and the tooth-specific axis data, thereby realizing the subject's tooth arrangement in an individually separated state as a 3D image and generating tooth arrangement data; a positional anomaly classification step of automatically classifying a plurality of positional anomalies including at least the shape of malocclusion by applying the relative positional anomalies of the tooth objects included in the tooth arrangement data generated in the image processing step, which are derived by comparing tooth data for the adjacent teeth based on the coordinates and the contact points of the representative spheres of each tooth object and the tooth-specific axis data, to an automatic diagnosis algorithm obtained by deep learning as a result of determining whether the relative positional anomalies correspond to at least one of crowding, spacing, rotation, perpendicular relationship (openbite & deepbite), mesiodistal inclination (tipping), buccolingual inclination (torque), and occlusion; and and a step of generating orthodontic data based on the result of the positional abnormality classification step by combining each tooth image, each tooth data, each tooth identification information, and the classification result for each tooth positional abnormality with the subject's pre-treatment oral cavity condition information.

Images