Method, apparatus, and system for providing automated solution for dental analysis and modeling based on 3D scanning data using artificial intelligence model

Abstract

One embodiment of the present invention relates to a technology that improves the efficiency of dental treatment processes by combining artificial intelligence (AI) technology and 3D image processing technology. More specifically, it relates to a method, device, and system for providing an automated solution for tooth analysis and modeling based on 3D scanning data using an artificial intelligence model, which automatically analyzes teeth and oral structures from 3D scanning data obtained through an oral scanner, and automates the entire process from extracting margin lines necessary for fabricating prostheses, analyzing clearance, and modeling the final prosthesis.

Description

Technology Field

[0001] The following embodiments relate to a technology that improves the efficiency of dental treatment processes by combining artificial intelligence (AI) technology and 3D image processing technology. More specifically, they relate to a method, device, and system for providing an automated solution for tooth analysis and modeling based on 3D scanning data using an artificial intelligence model, which automatically analyzes teeth and oral structures from 3D scanning data obtained through an oral scanner, and automates the entire process from extracting margin lines required for fabricating prostheses to analyzing clearance and modeling the final prosthesis. Background Technology

[0002] With the recent digitization of the dental treatment environment, the use of intraoral scanners and dental CAD software is becoming widespread.

[0003] The conventional analog method involved taking impressions of teeth and making plaster models, which left much room for patient discomfort and errors. However, the digital method replaced this with 3D data to increase precision.

[0004] However, the process of designing actual prostheses using acquired 3D scanning data still relies heavily on the manual work of skilled dental technicians.

[0005] For example, the dental technician must set the margin line, which is the boundary between the abutment tooth and the gum, by clicking it with a mouse on the scan data, and must modify the design while visually checking the occlusal relationship with the opposing tooth.

[0006] In particular, if the amount of tooth preparation performed by a dentist is insufficient to meet the minimum thickness required for prosthetic fabrication, problems may arise where design becomes impossible during the laboratory process or the fabricated prosthesis fractures easily.

[0007] In conventional technology, these problems were discovered only in the later stages of the design phase, resulting in wasted treatment time and costs, such as the need to remove teeth and re-scan.

[0008] In addition, there were technical limitations in automatically distinguishing the upper and lower jaws or mapping tooth numbers (dental formulas) due to noise in the scan data or unnecessary areas such as the palate and tongue. Prior art literature

[0009] (Patent Document 0001) KR 10-1878467 B (Patent Document 0002) KR 10-2528601 B (Patent Document 0003) KR 10-2657979 B The problem to be solved

[0010] The problem that one embodiment of the present invention aims to solve is to provide a method, apparatus, and system for providing an automated solution for tooth analysis and modeling based on 3D scanning data using an artificial intelligence model, which automates the preprocessing of 3D scan data, object recognition (segmentation), margin line identification, and upper and lower jaw differentiation through an artificial intelligence model in order to overcome the manual dependency limitations and inefficiencies of conventional dental CAD work as described above, and generates a prosthetic design in a one-stop manner based on this. means of solving the problem

[0011] According to one embodiment, a device includes a processor, memory, a communication module, and a non-transient storage medium, and a method for providing an automated solution for tooth analysis and modeling based on 3D scanning data utilizing an artificial intelligence model, wherein the processor executes a program stored in the non-transient storage medium, the method comprises: receiving first 3D scan data generated by scanning an oral region including a prepared tooth, which is the target of fabricating a prosthesis, in 3D, and second 3D scan data generated by scanning an oral region including an antagonist tooth in 3D; generating a segmentation result by processing the distinction between the upper and lower jaws, the separation of the tooth region and the gum region, and the recognition of dental notation corresponding to each tooth from the first 3D scan data using an artificial intelligence model; and identifying a margin line, which is the boundary between the prepared tooth and the gum, which is the target of fabricating a prosthesis, based on the segmentation result. A method for providing an automated tooth analysis and modeling solution based on 3D scanning data using an artificial intelligence model is provided, comprising: a step of registering the first 3D scan data and the second 3D scan data to determine whether the abutment tooth interferes with the opposing tooth and to calculate a separation distance secured for forming a prosthesis; and a step of automatically generating a 3D design model of a dental prosthesis to be mounted on the abutment tooth based on the margin line and the separation distance.

[0012] In addition, the step of calculating the separation distance comprises: identifying mesh regions constituting the occlusal surface, mesial surface, and distal surface of the abutment tooth, respectively, based on the segmentation results; extracting the surface mesh of the antagonist tooth and the surface mesh of the adjacent tooth adjacent to the abutment tooth, respectively; aligning the first 3D scan data and the second 3D scan data to align the abutment tooth and the antagonist tooth so that they are placed on the same coordinate system; measuring the projection distance from each vertex of the mesh constituting the occlusal surface of the abutment tooth to the surface of the antagonist tooth along the normal direction of each vertex of the occlusal surface, and determining the lower quantile value of the projection distances as the occlusal clearance. A step of calculating an occlusal plane by performing least-squares plane fitting on a set of vertices of a mesh constituting the occlusal surface of the abutment tooth; a step of generating a horizontal normal vector by projecting the normal vector of a set of lateral vertices, which are the vertices of the mesial and distal surfaces of the abutment tooth, onto the occlusal plane; a step of generating a ray in the direction of the horizontal normal vector to search for the nearest intersection point between the ray and the surface mesh of an adjacent tooth, calculating lateral distances which are the distances between each element of the set of lateral vertices and the nearest intersection point, and determining the lower quantile value of the lateral distances as the proximal clearance; a step of determining the minimum value among the occlusal clearance and the proximal clearance as the separation distance; a step of obtaining a preset minimum thickness threshold based on the material information of the prosthesis to be fabricated; a step of determining that if the separation distance is greater than or equal to the minimum thickness threshold, the state of being able to design a prosthesis based on the material is determined, and setting 'prosthesis minimum thickness ≥ minimum thickness threshold' as a design constraint for generating a 3D design model;and if the above separation distance is less than a minimum thickness threshold, the method may include the step of determining a material-based prosthesis design restriction state, outputting material candidate information for selecting an alternative material with a smaller minimum thickness threshold, or generating tooth reduction guide information including occlusal surface reduction amount, mesial surface reduction amount, and distal surface reduction amount.;

[0013] And, the step of generating the segmentation result comprises: a step of extracting a geometric feature vector including 3D position coordinates (x, y, z), normal vector, and curvature information for each of a plurality of vertices and faces constituting the preprocessed first 3D scan data; a step of inputting the geometric feature vector into an input layer of an artificial intelligence model; a step of calculating a class probability map indicating the probability that each vertex and face belongs to one of the Tooth Class, Gingiva Class, and Background Class through the operation of the artificial intelligence model; a step of performing instance segmentation by dividing the regions classified as tooth classes into individual tooth objects through connected component analysis and clustering based on the class probability map, and assigning an instance identifier to each individual tooth object; A step of deriving a result of distinguishing between the upper and lower jaws by analyzing the curvature of the dental arch and the presence of the palate region based on the class probability map, classifying whether the first 3D scan data corresponds to the maxilla or the mandible; a step of calculating centroid coordinates for each of the individual tooth objects and determining the tooth arrangement order on the entire dental arch using the centroid coordinates; a step of generating a dental notation recognition result by mapping a unique tooth number to each individual tooth object according to at least one of the FDI (World Dental Federation) notation and the Universal Numbering System based on the arrangement order on the dental arch and the result of distinguishing between the upper and lower jaws; and a step of designating the class probability map, instance identifier for each individual tooth object, result of distinguishing between the upper and lower jaws, tooth arrangement order, and dental notation recognition result as a segmentation result.It may include.;

[0014] In addition, the step of automatically generating the above 3D design model comprises: a step of obtaining a first standard tooth library model corresponding to the tooth recognition result from a database; a step of generating a second standard tooth library model that is initially placed at the location of the abutment tooth based on the centroid coordinates of the abutment tooth and the curvature of the dental arch, and scaled to correspond to the tooth size of the abutment tooth and the shape of the dental arch; a step of determining the external boundary of the prosthesis by performing mesh deformation on the second standard tooth library model so that the cervical line, which is the lower boundary line of the second standard tooth library model, coincides with the margin line; and a step of generating an intaglio surface mesh offset by a cement gap corresponding to a preset allowable adhesive thickness in the normal direction from the surface mesh of the abutment tooth. The method may include: a step of generating an outer surface mesh by performing shape optimization on the occlusal surface and lateral mesh of the second standard tooth library model to form occlusal points with opposing teeth and contact points with adjacent teeth so as to satisfy spacing distances and design constraints; and a step of combining the outer surface mesh and the inner mesh to generate a three-dimensional design model of a watertight structure having a margin line as a closed curve and an abutment tooth receiving space formed inside.

[0015] In addition, the step of deriving the upper and lower jaw distinction result comprises: a step of extracting an area classified as a gum class in the class probability map; a step of calculating a dental arch curve based on the dentition arrangement of the individual tooth objects; a step of establishing an inner region of the dental arch curve based on the dental arch curve and analyzing the connectivity and spatial occupancy of the gum class areas distributed in the inner region; and a step of identifying the inner region as continuous soft tissue constituting the palate and determining the first 3D scan data as the maxilla if, in the connectivity analysis result, the gum class mesh forms a single connectivity component and the spatial occupancy is greater than or equal to a preset occupancy threshold. and if the inner region is classified as a background class, or if the gum class mesh is separated into multiple connection components in the connectivity analysis result, or if the space occupancy is less than the occupancy threshold, the inner region is identified as a space corresponding to the tongue space or floor of mouth, and the first 3D scan data is determined to be the mandible; may be included.

[0016] A device according to one embodiment may be combined with hardware and controlled by a computer program stored on a medium to execute the method of any one of the methods described above. Effects of the invention

[0017] According to one embodiment, by automating a series of processes from receiving 3D scan data to preprocessing, segmentation, margin identification, and design generation through artificial intelligence, the time required for dental laboratory work can be drastically reduced and quality variations due to the skill level of the operator can be reduced.

[0018] Furthermore, the distance between the occlusal and lateral (proximal) surfaces of the abutment tooth and the opposing and adjacent teeth is precisely measured using projection distance and ray search techniques. By applying statistical methods (sub-quantiles) to eliminate errors caused by noise, the actual available clearance for prosthetic fabrication can be accurately determined. This allows for the pre-verification of design suitability based on prosthetic materials (zirconia, metal, etc.) and provides specific removal guidelines in case of insufficient reduction, thereby effectively preventing clinical failure.

[0019] Furthermore, through an AI model utilizing geometric feature vectors, teeth, gums, and the background can be precisely segmented. In particular, by applying unique logic that analyzes gingival connectivity and spatial occupancy in the medial region of the dental arch, the maxilla and mandible can be clearly distinguished without separate training for palatal classes. This enhances the learning efficiency of the AI ​​model while ensuring robust recognition performance in various oral environments (edentulous, partial tooth loss, etc.).

[0020] In addition, by scaling and morphing a standard tooth library according to the abutment tooth condition, creating an inner surface considering the cement gap and optimizing the outer surface considering the occlusal / adjacent relationship, and then combining them into a watertight structure, high-quality prosthetic design data that can be immediately 3D printed or milled without separate post-modification can be automatically generated. Brief explanation of the drawing

[0021] FIG. 1 is a schematic diagram showing a system for providing an automated solution for tooth analysis and modeling based on 3D scanning data using an artificial intelligence model according to an embodiment of the present invention. FIG. 2 is a flowchart illustrating a method for providing an automated tooth analysis and modeling solution based on 3D scanning data using an artificial intelligence model according to an embodiment of the present invention. FIG. 3 is a flowchart illustrating the step of calculating the separation distance of a method for providing an automated solution for tooth analysis and modeling based on 3D scanning data using an artificial intelligence model according to an embodiment of the present invention. FIG. 4 is a flowchart illustrating the steps for generating segmentation results of a method for providing an automated tooth analysis and modeling solution based on 3D scanning data using an artificial intelligence model according to an embodiment of the present invention. FIG. 5 is a flowchart illustrating the steps of automatically generating a three-dimensional design model of a method for providing an automated solution for tooth analysis and modeling based on 3D scanning data using an artificial intelligence model according to an embodiment of the present invention. Specific details for implementing the invention

[0022] Hereinafter, embodiments are described in detail with reference to the attached drawings. However, various modifications may be made to the embodiments, and thus the scope of the patent application is not limited or restricted by these embodiments. It should be understood that all modifications, equivalents, and substitutions to the embodiments are included within the scope of the rights.

[0023] Specific structural or functional descriptions of the embodiments are disclosed for illustrative purposes only and may be modified and implemented in various forms. Accordingly, the embodiments are not limited to the specific disclosed forms, and the scope of this specification includes modifications, equivalents, or substitutions that fall within the technical concept.

[0024] Terms such as "first" or "second" may be used to describe various components, but these terms should be interpreted solely for the purpose of distinguishing one component from another. For example, the first component may be named the second component, and similarly, the second component may be named the first component.

[0025] When it is stated that a component is "connected" to another component, it should be understood that it may be directly connected to or coupled with that other component, or that there may be other components in between.

[0026] The terms used in the embodiments are for illustrative purposes only and should not be interpreted as intended to be limiting. Singular expressions include plural expressions unless the context clearly indicates otherwise. In this specification, terms such as "comprising" or "having" are intended to indicate the existence of the features, numbers, steps, actions, components, parts, or combinations thereof described in the specification, and should be understood as not precluding the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof.

[0027] In particular, where a 'step' in this specification is described as 'comprising' one or more detailed steps or sub-steps, said 'step' may be interpreted as including its own basic processing step while simultaneously performing the described detailed steps as well.

[0028] For example, if it is stated that 'a step of doing B to A' includes 'a step of doing D to C; a step of doing F to E; and a step of doing H to G,' the 'step of doing B to A' may be interpreted not merely as the basic operation of doing B to A, but as a configuration that performs detailed procedures together, such as a step of doing D to C, a step of doing F to E, and a step of doing H to G.

[0029] Accordingly, the above configuration does not exclude various sub-procedures included within the scope of execution of the corresponding step, and may be included within the scope of the present invention even if other procedures or means performing substantially the same or equivalent functions are substituted.

[0030] Expressions such as 'end part', 'both ends', 'one end', 'other end', and 'side end' of a component can be interpreted as referring to at least / any one of the end parts of that component.

[0031] In the description of the present invention, 'a method in which a device comprises a processor, a memory, a communication module, and a non-transient storage medium, and a program stored in the non-transient storage medium is executed by the processor,' the term 'method' may be interpreted as referring to the program stored in the non-transient storage medium itself or a part of the program.

[0032] The term 'Return' as used in the description of the present invention may refer to a result value being output, returned, or returned from a method, procedure, function, etc. used in a given program language / structure.

[0033] Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as generally understood by those skilled in the art to which the embodiments pertain. Terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and should not be interpreted in an ideal or overly formal sense unless explicitly defined in this application. For example, the term 'artificial intelligence model' may be selected from one or more of known general artificial intelligence models.

[0034] In addition, when describing with reference to the attached drawings, identical components are assigned the same reference numeral regardless of drawing symbols, and redundant descriptions thereof are omitted. In describing the embodiments, if it is determined that a detailed description of related prior art could unnecessarily obscure the essence of the embodiments, such detailed description is omitted.

[0035] According to one embodiment, a device includes a processor, memory, a communication module, and a non-transient storage medium, and a method for providing an automated solution for tooth analysis and modeling based on 3D scanning data utilizing an artificial intelligence model, wherein the processor executes a program stored in the non-transient storage medium, the method comprises: receiving first 3D scan data generated by scanning an oral region including a prepared tooth, which is the target of fabricating a prosthesis, in 3D, and second 3D scan data generated by scanning an oral region including an antagonist tooth in 3D; generating a segmentation result by processing the distinction between the upper and lower jaws, the separation of the tooth region and the gum region, and the recognition of dental notation corresponding to each tooth from the first 3D scan data using an artificial intelligence model; and identifying a margin line, which is the boundary between the prepared tooth and the gum, which is the target of fabricating a prosthesis, based on the segmentation result. A method for providing an automated tooth analysis and modeling solution based on 3D scanning data using an artificial intelligence model is provided, comprising: a step of registering the first 3D scan data and the second 3D scan data to determine whether the abutment tooth interferes with the opposing tooth and to calculate a separation distance secured for forming a prosthesis; and a step of automatically generating a 3D design model of a dental prosthesis to be mounted on the abutment tooth based on the margin line and the separation distance.

[0036] In the step of receiving first 3D scan data generated by scanning an oral region including a prepared tooth, which is the target of fabricating a prosthesis, and second 3D scan data generated by scanning an oral region including an antagonist tooth, the data generated by 3D scanning of the oral region including the prepared tooth and the oral region including the antagonist tooth is received / input from an external device.

[0037] At this time, the first 3D scan data and the second 3D scan data may be received / inputted in a form stored in a single integrated file, or may be received / inputted in a separate form.

[0038] Here, the first three-dimensional scan data refers to shape data of a jaw (upper or lower jaw) containing an abutment tooth in which the outer surface of the tooth has been prepared for prosthetic treatment.

[0039] A prepared tooth refers to a tooth that has been reduced in volume by a dentist in order to place a prosthetic device, such as a crown or a bridge.

[0040] The first three-dimensional scan data may include a mesh structure composed of multiple vertices and faces, a point cloud structure, or a combination thereof, and each vertex may include three-dimensional position coordinates, and in some cases, normal vectors, colors, scan quality indicators (e.g., reliability) may be added.

[0041] The first three-dimensional scan data may be oral data of a patient obtained through an intraoral scanner or a model scanner, and may be received in a file format (e.g., STL, OBJ, PLY, etc.) or sequentially received from a scanner in a streaming format, and in the case of sequential reception, the mesh accumulated in units of a certain interval may be merged to be determined as one first three-dimensional scan data.

[0042] Opposing teeth refer to teeth that come into contact with or face the abutment teeth when chewing food (occlusion), and mean teeth of the opposite arch that interlock with the abutment teeth to form an occlusal relationship.

[0043] When fabricating a prosthesis, the height and shape must be designed considering the occlusal relationship with the opposing tooth, so opposing tooth data (second 3D scan data) is required separately from abutment tooth data (first 3D scan data).

[0044] The above second three-dimensional scan data may be configured to include a range comprising the occlusal surface of the opposing tooth and the surrounding dentition.

[0045] Since the first 3D scan data and the second 3D scan data can be acquired in different coordinate systems, reference information (e.g., scan time, scanner coordinate system, feature point, byte scan, etc.) for aligning to the same coordinate system in the subsequent alignment step may be received together, or separate byte relationship data may be additionally received.

[0046] In some cases, pre-configured standard model data may be used as second three-dimensional data instead of actual opposing tooth scan data.

[0047] In the step of performing preprocessing including at least one of hole filling, noise removal, surface smoothing, and shape correction on the mesh structure of the first three-dimensional scan data, at least one of hole filling, noise removal, surface smoothing, and shape correction is performed on the mesh structure of the first three-dimensional scan data to secure shape quality suitable for subsequent analysis.

[0048] Preprocessing is a process of refining defects or unnecessary data included in the original scan data and processing it into a state that can be analyzed. Depending on the embodiment, after completing post-processing (preprocessing from the perspective of the device) on the 3D scanner itself, data in a form that can be immediately input into an artificial intelligence model, etc. may be received from the device (server).

[0049] During the oral scanning process, holes or noise may occur in the data due to interference from soft tissues such as saliva, tongue, and cheeks, light reflection, or angle constraints of the scanner.

[0050] Hole filling refers to the process of naturally filling in gaps caused by missing data into a closed surface based on the curvature or normal vector information of the surrounding mesh. Specifically, this can be performed by detecting the closed curves of the hole boundaries, filling the interior of the boundaries using triangulation, and then repositioning the vertices to align with the surrounding surface normal directions.

[0051] Noise removal refers to the process of removing floating noise or spike-like noise that roughens the surface, which is suspended in space rather than on the surface of the teeth or gums. Specifically, this can be performed by removing isolated connection components below a certain area, or by identifying and removing areas where the local curvature of a vertex spikes sharply as outliers.

[0052] Surface smoothing refers to the process of smoothing out fine irregularities on a mesh surface by applying algorithms such as Laplacian smoothing. It can be applied restrictively to mitigate high-frequency shape noise without excessive loss of geometric features near tooth boundaries or margins.

[0053] Shape correction refers to the process of correcting distorted or overlapping mesh surfaces to match normal anatomical shapes due to scanner coordinate system distortion, partial scan merging errors, hand tremors, etc.

[0054] The first 3D scan data preprocessed in this way serves as a basis for reducing false detections in subsequent AI model-based segmentation and matching, and for increasing the stability of distance calculation and shape optimization.

[0055] In the step of using an artificial intelligence model to distinguish the upper and lower jaws from the preprocessed first 3D scan data, separating the tooth region and the gum region, and recognizing the dental notation corresponding to each tooth to generate a segmentation result, a deep learning-based artificial intelligence model is used to semantically segment what each part of the 3D data is.

[0056] The artificial intelligence model may be defined as a classification model that takes a 3D mesh or point cloud as input and outputs a class probability at the vertex / face level, an instance partitioning model that outputs an instance identifier per tooth object, or a combination thereof. Specifically, it may include a neural network structure capable of processing 3D point cloud or mesh data, such as PointNet, PointNet++, DGCNN, or MeshSegNet.

[0057] Distinguishing between the maxilla and mandible involves determining whether the input 1st 3D scan data is the maxilla or the mandible, and may include a procedure for determining whether the input scan is the maxilla or the mandible based on the same tooth / gum segmentation results. For example, this can be performed by analyzing the presence of a palate region, the shape of the dental arch, the overall directionality of the normal vector, etc.

[0058] Separation of tooth and gum regions refers to classifying whether each vertex or face constituting the mesh belongs to the hard tissue 'teeth' or the soft tissue 'gums'.

[0059] Dental notation recognition refers to the process of identifying the anatomical tooth number of each segmented tooth object (e.g., #11 for the maxillary right central incisor, #36 for the mandibular left first molar, etc.) and assigning a unique number. This can be defined as the process of determining the arrangement order of individual tooth objects and mapping tooth numbers according to standard notations (such as FDI notation or the Universal Numbering System).

[0060] It is desirable that the segmentation results generated in this step be configured to include at least class information regarding teeth, gums, and background, instance identifiers of individual tooth objects, results of upper and lower jaw differentiation, and results of tooth formula recognition. The accuracy of this process directly affects the quality of subsequent standard tooth library model selection / initial placement and prosthetic shape generation.

[0061] In the step of identifying the margin line, which is the boundary between the abutment tooth and the gum to be fabricated, based on the above segmentation results, the margin line corresponding to the boundary between the abutment tooth and the gum is automatically identified based on the above segmentation results.

[0062] The margin line is the boundary where the tip of the prepared abutment tooth meets the gum; it can be defined as the boundary around the cervical region where the lower edge of the prosthesis contacts the abutment tooth when the prosthesis is mounted. It is the most important reference line for ensuring that the inner surface of the prosthesis adheres precisely to the tooth.

[0063] The margin line can be implemented as a closed or semi-closed curve formed along the boundary area where the surface mesh of the abutment tooth and the surface mesh of the gingival area meet.

[0064] In this step, the boundary area where the 'tooth class' and 'gum class' meet is primarily extracted from the previously performed segmentation results, and a closed curve-shaped margin line can be generated by connecting feature points within that area where the curvature of the mesh changes abruptly.

[0065] Specifically, a set of boundary vertices located at the boundary with vertices classified as gum class among mesh vertices classified as abutment teeth is extracted, and a continuous polyline is generated by aligning the set of boundary vertices in the direction of the dental arch, and then curve smoothing and outlier removal are performed to determine the final margin line.

[0066] In cases where margin lines are extracted discontinuously (e.g., classification errors due to scan defects or hemorrhage / reflection), the gap between adjacent sections can be connected through interpolation if the gap is below a preset connection threshold, or corrected by prioritizing the selection of the ridge area with the largest change in the abutment normal direction and curvature.

[0067] The margin lines identified in this way can be utilized as boundary conditions in the subsequent generation of the prosthesis inner mesh and the generation of a watertight structure model.

[0068] In the step of registering the first three-dimensional scan data and the second three-dimensional scan data to determine whether the abutment tooth interferes with the opposing tooth and to calculate the separation distance secured for forming the prosthesis, the first three-dimensional scan data and the second three-dimensional scan data are registered to align the abutment tooth and the opposing tooth so that they are placed in the same coordinate system, then the abutment tooth interferes with the opposing tooth and to calculate the separation distance secured for forming the prosthesis.

[0069] Registration refers to the process of aligning first 3D scan data and second 3D scan data, which have different coordinate systems, on a single common coordinate system so that they have the same positional relationship as the actual patient with their mouth closed (occlusion state).

[0070] To this end, separate bite scan data can be used as a medium, or known 3D alignment techniques such as the ICP (Iterative Closest Point) algorithm and feature point-based initial alignment can be used.

[0071] The determination of interference involves checking whether the surface of the abutment tooth penetrates (collides with) the surface of the opposing tooth in the aligned state. Specifically, this can be determined based on whether the minimum distance between the surface of the abutment tooth and the surface of the opposing tooth after alignment is 0 or less, or whether there exists an area below a preset interference threshold.

[0072] Separation distance refers to the shortest physical distance from the surface of the abutment tooth to the surface of the opposing tooth, and corresponds to the clearance, which is the space where the prosthesis can be placed. Separation distance can be determined as the minimum value between the occlusal gap and the adjacent gap.

[0073] For example, if the prosthetic material is zirconia and a minimum thickness of 1.0 millimeter (mm) is required, but the calculated spacing is 0.8 millimeters, it may be judged as 'insufficient spacing'.

[0074] In this step, a processing flow that reliably calculates the “distance reflecting the occlusal direction” and the “distance reflecting the adjacent direction” using the matched results can be utilized as a key element.

[0075] After calculating the separation distance, a designable or design-limited state can be determined by reflecting a minimum thickness threshold based on the material of the prosthesis to be fabricated; in the case of a design-limited state, the possibility of prosthesis fabrication failure or excessive remanufacturing can be reduced by outputting alternative material candidate information or generating tooth removal guide information.

[0076] In the step of automatically generating a three-dimensional design model of a dental prosthesis to be mounted on an abutment tooth based on the above margin line and spacing distance, the shape of the prosthesis is automatically modeled by synthesizing the analyzed information.

[0077] Based on the above margin line and spacing, a 3D design model of a dental prosthesis to be mounted on an abutment tooth is automatically generated.

[0078] A dental prosthesis is an artificial replacement fabricated to restore a damaged tooth, and may include crowns, inlays, onlays, veneers, etc.

[0079] A three-dimensional design model can be defined as a watertight structure prosthetic mesh including an outer mesh and an inner mesh.

[0080] The process of automatically generating a 3D design model may include the process of loading a standard tooth library model that matches the tooth formula of the abutment tooth (initial placement and scaling), the process of morphing the bottom of the model to match the identified margin line to determine the external boundary, and the process of optimizing the height and surface of the model to form an appropriate occlusal point without contacting the opposing tooth while considering the calculated separation distance (shape optimization).

[0081] The inner mesh can be created with a shape offset from the abutment tooth surface by a preset cement gap.

[0082] The outer mesh can be generated by performing shape optimization for forming occlusal points and contact points to satisfy the spacing distance and design constraints in relation to the matched opposing teeth and adjacent teeth.

[0083] Finally, the device combines the outer mesh and the inner mesh, maintaining the margin line as a closed curve and forming an abutment tooth reception space inside, thereby producing a 3D design model of a watertight prosthesis structure required in the manufacturing process.

[0084] The generated 3D design model is saved in a file format (e.g., STL) that can be immediately produced on a milling machine or 3D printer and provided to the user.

[0085] In addition, the step of calculating the separation distance comprises: identifying mesh regions constituting the occlusal surface, mesial surface, and distal surface of the abutment tooth, respectively, based on the segmentation results; extracting the surface mesh of the antagonist tooth and the surface mesh of the adjacent tooth adjacent to the abutment tooth, respectively; aligning the first 3D scan data and the second 3D scan data to align the abutment tooth and the antagonist tooth so that they are placed on the same coordinate system; measuring the projection distance from each vertex of the mesh constituting the occlusal surface of the abutment tooth to the surface of the antagonist tooth along the normal direction of each vertex of the occlusal surface, and determining the lower quantile value of the projection distances as the occlusal clearance. A step of calculating an occlusal plane by performing least-squares plane fitting on a set of vertices of a mesh constituting the occlusal surface of the abutment tooth; a step of generating a horizontal normal vector by projecting the normal vector of a set of lateral vertices, which are the vertices of the mesial and distal surfaces of the abutment tooth, onto the occlusal plane; a step of generating a ray in the direction of the horizontal normal vector to search for the nearest intersection point between the ray and the surface mesh of an adjacent tooth, calculating lateral distances which are the distances between each element of the set of lateral vertices and the nearest intersection point, and determining the lower quantile value of the lateral distances as the proximal clearance; a step of determining the minimum value among the occlusal clearance and the proximal clearance as the separation distance; a step of obtaining a preset minimum thickness threshold based on the material information of the prosthesis to be fabricated; a step of determining that if the separation distance is greater than or equal to the minimum thickness threshold, the state of being able to design a prosthesis based on the material is determined, and setting 'prosthesis minimum thickness ≥ minimum thickness threshold' as a design constraint for generating a 3D design model;and if the above separation distance is less than a minimum thickness threshold, the method may include the step of determining a material-based prosthesis design restriction state, outputting material candidate information for selecting an alternative material with a smaller minimum thickness threshold, or generating tooth reduction guide information including occlusal surface reduction amount, mesial surface reduction amount, and distal surface reduction amount.;

[0086] The step of calculating the above separation distance is to quantify the occlusal distance between the abutment tooth and the opposing tooth and the lateral distance between the abutment tooth and the adjacent tooth, respectively, using the segmentation results and the aligned first three-dimensional scan data and second three-dimensional scan data, and to determine the minimum distance that is more restrictive from the perspective of prosthetic design among the above distances as the separation distance, and then to determine whether design is possible based on the result of comparison with the minimum thickness threshold according to the fabrication material, and to generate alternative material candidates or tooth reduction guides if necessary.

[0087] Here, “separation distance” can be defined as a value representing the minimum space that the outer surface of the prosthesis must secure, and the unit can be standardized to a length unit commonly used in dental CAD, such as millimeters.

[0088] Based on the above segmentation results, the step of identifying mesh regions constituting the occlusal, mesial, and distal surfaces of the abutment tooth, respectively, involves first extracting an abutment tooth mesh corresponding to an abutment tooth instance identifier included in the above segmentation results, then distinguishing the occlusal surface region and the lateral region based on the vertices and faces of the abutment tooth mesh according to the “face direction” and “relative position within the dental arch,” and subdividing the lateral region into the mesial and distal surfaces.

[0089] The occlusal surface refers to the upper surface of the tooth that occludes with the opposing tooth, the mesial surface refers to the lateral surface located anteriorly (towards the midline) along the dental arch, and the distal surface refers to the lateral surface located posteriorly (towards the wisdom tooth). The reason for distinguishing these regions within the entire mesh of the abutment tooth is to measure distances in different ways, as the vertical thickness (height) of the occlusal surface is important for prosthetic fabrication, while the horizontal distance (contact point) from adjacent teeth is important for the mesial and distal surfaces.

[0090] Specifically, the device can identify each face by clustering the normal vector direction or position coordinates (such as Z-axis height) of each vertex predicted by an artificial intelligence model. For example, vertices in the abutment mesh whose normal vectors have a strong occlusal direction component (e.g., up-down direction) can be classified as occlusal surface candidates, while vertices with a weak occlusal direction component and a strong lateral component can be classified as lateral surface candidates.

[0091] To distinguish between the mesial and distal surfaces, the device establishes the anterior and posterior directions of the dentition based on the coordinates of the dental arch curve or the center of gravity of the abutment tooth, and classifies the lateral candidate vertex as the mesial surface if located on the anterior side and the distal surface if located on the posterior side. At this time, the area near the boundary vertex is designated as a buffer zone to ensure that the occlusal, mesial, and distal surfaces do not overlap, and the vertex of the buffer zone can be excluded from subsequent distance calculations or have its weight reduced to ensure calculation stability.

[0092] The step of extracting the surface mesh of the opposing tooth and the surface mesh of the adjacent tooth adjacent to the abutment tooth, respectively, is a step of defining a counterpart to be measured for distance, wherein the tooth mesh corresponding to the opposing tooth is extracted from the second 3D scan data, and the adjacent tooth mesh that is in an adjacent relationship with the abutment tooth is extracted from the first 3D scan data.

[0093] The mesh on the surface of the opposing tooth serves as a reference for measuring the vertical separation distance (occlusal gap), and the mesh on the surface of the adjacent tooth serves as a reference for measuring the horizontal separation distance (adjacency gap).

[0094] Here, “opposing teeth” include teeth or dentition of the opposite arch that form an occlusal relationship with the abutment tooth, and “adjacent teeth” may be defined as teeth that have the potential to form a point of contact (where the potential is greater than or equal to a critical probability) on the mesial or distal side of the abutment tooth.

[0095] The device can identify the tooth number of the abutment tooth using the tooth recognition result of the segmentation result, and set tooth objects corresponding to the adjacent numbers on the left and right sides of the dentition for the said abutment tooth number as adjacent tooth candidates.

[0096] Considering the possibility of missing actual scan data or errors in tooth recognition, the device may further include a correction procedure to determine as adjacent teeth a tooth object that is located within a certain radius based on the center of gravity of the abutment tooth mesh and has the smallest minimum distance from the abutment tooth mesh.

[0097] At this time, the extracted mesh can be cropped to the area of ​​the opposing tooth physically closest to the abutment tooth and the adjacent tooth areas on both sides within the entire oral scan data to increase computation speed.

[0098] The step of aligning the first three-dimensional scan data and the second three-dimensional scan data so that the abutment tooth and the opposing tooth are placed on the same coordinate system is a step of unifying two data having different scan coordinate systems into one World Coordinate System.

[0099] A matching transformation is calculated and applied so that the first three-dimensional scan data and the second three-dimensional scan data acquired in different coordinate systems can be used to calculate distance in the same reference coordinate system. This is a process identical to or an extension of the matching step in claim 1 above, and is an essential prerequisite for accurate distance measurement.

[0100] The device can minimize residual error by performing initial alignment using byte scan data or common feature points (e.g., occlusal surface features, cusp vertices, ridge lines, etc.) and then performing vertex-vertex or vertex-plane-based iterative Closest Point (ICP).

[0101] Since the distance may be underestimated or overestimated if the quality of the alignment is insufficient, the device can verify the alignment quality by calculating the residual distribution after alignment (e.g., the mean and upper quantiles of the nearest point distance), and if the quality does not satisfy the threshold criteria, it can re-perform the alignment or output correction request information to the administrator terminal.

[0102] The step of measuring the projected distance from each vertex of the mesh constituting the occlusal surface of the abutment tooth to the surface of the opposing tooth along the normal direction of each vertex of the occlusal surface, and determining the lower quantile values ​​of the projected distances as occlusal intervals (or in the step of determining the occlusal intervals) is to calculate the “occlusal direction distance” to the opposing tooth for each vertex of the abutment occlusal surface in a matched coordinate system, and to determine the lower quantile values ​​of the distances as representative intervals in order to reduce the influence of extreme values.

[0103] The reason for following the 'normal direction' rather than simply measuring the shortest distance is that the direction in which the thickness of the prosthesis is formed is perpendicular to the tooth surface.

[0104] “Projection distance” can be defined as the distance from each vertex to the point where a virtual line segment or ray is advanced in the direction of the vertex's normal to first intersect the opposing tooth surface mesh.

[0105] “Lower quantile value” is a value representing the lower part of the distance distribution, and can be set, for example, as the lower quartile or lower decile (e.g., the value at the lower 5% or 10% point), and can be used as a value to reliably represent the closest possible contact area among the entire occlusal surface.

[0106] Even if the distance of a specific point is measured to be close to 0 due to noise in the scan data, setting this as the representative value of the whole will cause an error, so the lower quantile value is used to exclude noise and safely secure the actual minimum thickness.

[0107] For example, when the projected distances calculated from the occlusal plane vertices are distributed as approximately 2 millimeters, 1.7 millimeters, 1.6 millimeters, 1.5 millimeters, etc., the device can determine the occlusal interval by fixing the lower quantile value to the 1.5 millimeter level.

[0108] If a vertex that does not intersect the opposing tooth exists during the projection process (e.g., the normal is in the opposite direction, or the ray escapes into the air due to an opening), the device can obtain a valid distance sample by invalidating the distance of that vertex or by reprojecting using a reference direction corrected in the direction of the occlusal plane.

[0109] The step of calculating the occlusal plane by performing least squares plane fitting on the set of vertices of the mesh constituting the occlusal surface of the abutment tooth is a step of estimating the plane that best fits the three-dimensional distribution of the occlusal surface vertices in order to define the overall inclination and reference direction of the abutment tooth occlusal surface.

[0110] Since teeth are composed of curved surfaces, the definition of 'horizontal' can be ambiguous; therefore, an occlusal plane—a virtual plane (fitting plane) that statistically best describes the points on the occlusal surface of the abutment tooth—is calculated and used as the standard for horizontality.

[0111] Here, “least squares plane fitting” can be defined as a process for determining the plane normal and position such that the sum of squares of the degree of deviation of vertices from the plane is minimized, and can stably reflect the overall trend even if there is some defect or noise in the occlusal surface.

[0112] The device can reduce distortion of the occlusal plane by identifying protruding vertices with excessive curvature or preprocessed residual noise vertices among the occlusal surface vertices as outliers and excluding them from the fitting target. The calculated occlusal plane can be used as a reference plane in the subsequent generation of horizontal normal vectors.

[0113] The step of generating a horizontal normal vector by projecting the normal vector of a set of lateral vertices, which are the vertices of the mesial and distal surfaces of the abutment tooth, onto the occlusal plane is a step of removing the normal direction component of the occlusal plane from the normal vector of the lateral vertices and leaving only the component within the occlusal plane in order to measure the lateral spacing as a “direction component parallel to the occlusal plane.”

[0114] The normal vector on the lateral side of the abutment tooth may be tilted upward or downward, but the distance to adjacent teeth must be measured primarily in the horizontal direction to secure an accurate path of insertion for the prosthesis. Therefore, the normal vector of the lateral vertex is projected onto the previously calculated occlusal plane to generate a horizontal normal vector with the tilt removed.

[0115] Here, the “horizontal normal vector” can be defined as the reference direction of the ray proceeding from the lateral vertex toward the adjacent tooth, and even if the original normal vector of the lateral vertex includes upper and lower components, the calculation of the lateral distance can be configured to be less sensitive to changes in the occlusal direction by removing the upper and lower components.

[0116] If the magnitude of the vector after projection is excessively small and the directionality is unstable (e.g., a vertex with a nearly vertical normal), the vertex can be excluded from the set of lateral vertices, or an alternative direction can be generated by averaging the horizontal normal vectors of adjacent vertices.

[0117] The step of generating a ray in the direction of the horizontal normal vector to search for the nearest intersection point between the ray and the surface mesh of the adjacent tooth, calculating lateral distances which are the distances between each element of the lateral vertex set and the nearest intersection point, and determining the sub-quantile values ​​of the lateral distances as adjacent intervals is a step of measuring the intersection distance with the adjacent tooth surface for each lateral vertex and determining a representative value as a sub-quantile in order to quantify the minimum possible contact space from the side of the abutment tooth to the adjacent tooth.

[0118] A virtual ray is fired in the direction of the horizontal normal vector to measure the distance to the point where it meets an adjacent tooth. Here, the “ray” can be defined as a ray or a finite-length line segment starting from a lateral vertex and proceeding in the direction of the horizontal normal vector, and the “nearest intersection point” can refer to the intersection point among the points where the ray intersects the adjacent tooth surface mesh that is the shortest distance from the starting point.

[0119] When adjacent tooth meshes exist on the mesial and distal sides, respectively, the device can ensure that the definition of lateral distances aligns with the dentition direction by first searching for intersections with the mesial adjacent tooth mesh for the mesial vertex set and first searching for intersections with the distal adjacent tooth mesh for the distal vertex set.

[0120] In this case as well, to prevent measurement errors caused by noise, the lower quantile value among the multiple measured lateral distance values ​​is determined as the adjacent interval. Using the lower quantile value of the calculated lateral distances as a representative value is intended to mitigate over-distance (e.g., when the ray slips out of the air, abnormal intersection due to mesh defects) or under-distance (e.g., intersection with noise fragments of adjacent tooth mesh) that may occur at some vertices. This allows for the identification of the practical minimum width at which the prosthesis can fit without getting stuck between adjacent teeth.

[0121] The step of determining the minimum value among the occlusal gap and adjacent gap as the separation distance is a step of determining the smaller value between the occlusal gap and adjacent gap as the separation distance in order to use the gap that constitutes the most restrictive spatial constraint in the prosthesis design as the representative value.

[0122] This is to determine the feasibility of the design based on the weakest part (which will be made the thinnest) during the fabrication of the prosthesis.

[0123] For example, if the occlusal gap is one millimeter and the adjacent gap is 0.8 millimeters, the device determines the separation distance as 0.8 millimeters and can subsequently reflect this in the optimization of the prosthesis's outer surface shape and the determination of design feasibility.

[0124] If either the occlusal gap or the adjacent gap cannot be calculated due to an insufficient number of valid samples, the device may use another calculable gap as a temporary separation distance and assign a “uncertain” status flag to prompt additional scanning or administrator review in subsequent steps.

[0125] The step of obtaining a preset minimum thickness threshold based on material information of the prosthesis to be manufactured (or the step of obtaining the minimum thickness threshold) is a step of obtaining a minimum thickness criterion required to satisfy manufacturability and fracture strength requirements for each prosthesis material from a database or lookup table.

[0126] Depending on the material of the prosthesis, the minimum thickness that can be withstood without fracture varies. For example, gold alloys have good ductility and can be manufactured thinly (e.g., 0.5 mm), but zirconia or ceramics are brittle, so a thickness of at least a certain level (e.g., 1.0 mm) is essential. A threshold value suitable for the material selected by the user is obtained.

[0127] “Material information” may include identifiers, grades, and manufacturing process parameters (e.g., sintering shrinkage rate, milling tool diameter limits) of materials usable for fabricating prostheses, such as zirconia, metal alloys, and resin series.

[0128] The “minimum thickness threshold” can be defined as the recommended minimum thickness for each material, and since it can be set to different values ​​depending on the type of prosthesis (e.g., crown, inlay) or manufacturing equipment conditions even for the same material, the device can determine the threshold by referring to the prosthesis type and manufacturing equipment profile together with the material identifier.

[0129] When the above separation distance is greater than or equal to the minimum thickness threshold, the step of determining that a material-based prosthesis design is possible and setting the minimum thickness of the prosthesis as a design constraint to be greater than or equal to the minimum thickness threshold for generating a 3D design model is a step of determining that the prosthesis shape can be created with the current material because the secured separation distance satisfies the required thickness of the material, and registering a constraint so that the prosthesis thickness does not fall below the threshold in subsequent shape optimization.

[0130] If it is determined that sufficient space has been secured, the system is controlled to proceed to the next step, automatic design generation. At this stage, when the AI ​​model or design algorithm generates the prosthesis, constraints are set to prevent it from being made thinner than a specified threshold, thereby ensuring the durability of the prosthesis.

[0131] For example, when optimizing the outer mesh of a prosthesis, if the thickness between the inner mesh and the outer mesh at a specific location is about to become smaller than a threshold, the device can control the device to prioritize satisfying the thickness constraint by moving the outer surface at that location outward within a range that avoids interference with opposing / adjacent teeth, or by adjusting the target weight for contact point formation.

[0132] The device stores the separation distance, occlusal gap, adjacent gap, and applied material threshold values ​​along with the design feasibility determination results, which can be utilized as a basis for future manufacturing reports and verification.

[0133] If the above separation distance is less than the minimum thickness threshold, the step of determining a material-based prosthesis design limitation state, outputting material candidate information for selecting an alternative material with a smaller minimum thickness threshold, or generating tooth reduction guide information including occlusal surface reduction amount, mesial surface reduction amount, and distal surface reduction amount is a step of determining that there is insufficient space for forming a prosthesis that satisfies the required minimum thickness with the current material, and providing an alternative path to resolve the deficiency.

[0134] Inform the user of the risk of fracture during prosthesis fabrication due to insufficient space and present a solution.

[0135] Solution 1 is 'Material Change,' which recommends high-strength materials (e.g., metal) that can be fabricated in the current space. The “Material Candidate Information Output” may include a list of materials with a lower minimum thickness threshold than the current material, the threshold for each material, expected strength / aesthetic properties, and suitability for the fabrication process.

[0136] Solution 2 is 'Prep Guide,' which generates and provides guide information that visually indicates the location and amount to be removed so that the dentist can remove more of the abutment tooth to fill the insufficient space (e.g., if the current amount is 0.8mm and the required amount is 1.0mm, an additional 0.2mm needs to be removed).

[0137] “Tooth reduction guide information” can be defined as information that indicates how much additional reduction is required on which side of the abutment tooth to secure the insufficient thickness, and the occlusal, mesial, and distal reduction amounts can be calculated by reflecting the occlusal and lateral gap deficiencies, respectively.

[0138] For example, the device can calculate the difference between the minimum thickness threshold and the separation distance as the “additional distance required” and determine the amount of removal by distributing the said additional distance to the occlusal surface and the lateral surface. In this case, the distribution rule can be set such that if the occlusal gap is more restrictive, the weight of the occlusal surface removal is increased, and if the adjacent gap is more restrictive, the weight of the mesial or distal surface removal is increased.

[0139] To prevent the removal guide from being excessively presented, the device stores a clinically acceptable maximum removal range by material / tooth type and can output an auxiliary message, such as “additional scan required” or “recommend material change first,” if the calculated removal amount exceeds the maximum range.

[0140] Furthermore, the device can convert the removal guide into three-dimensional visualization information (e.g., color map on the abutment mesh, highlighting of the removal recommendation area) and provide it to an administrator terminal or dental CAD terminal, thereby enabling the location and direction of the removal area to be clearly conveyed.

[0141] And, the step of generating the segmentation result comprises: a step of extracting a geometric feature vector including 3D position coordinates (x, y, z), normal vector, and curvature information for each of a plurality of vertices and faces constituting the preprocessed first 3D scan data; a step of inputting the geometric feature vector into an input layer of an artificial intelligence model; a step of calculating a class probability map indicating the probability that each vertex and face belongs to one of the Tooth Class, Gingiva Class, and Background Class through the operation of the artificial intelligence model; a step of performing instance segmentation by dividing the regions classified as tooth classes into individual tooth objects through connected component analysis and clustering based on the class probability map, and assigning an instance identifier to each individual tooth object; A step of deriving a result of distinguishing between the upper and lower jaws by analyzing the curvature of the dental arch and the presence of the palate region based on the class probability map, classifying whether the first 3D scan data corresponds to the maxilla or the mandible; a step of calculating centroid coordinates for each of the individual tooth objects and determining the tooth arrangement order on the entire dental arch using the centroid coordinates; a step of generating a dental notation recognition result by mapping a unique tooth number to each individual tooth object according to at least one of the FDI (World Dental Federation) notation and the Universal Numbering System based on the arrangement order on the dental arch and the result of distinguishing between the upper and lower jaws; and a step of designating the class probability map, instance identifier for each individual tooth object, result of distinguishing between the upper and lower jaws, tooth arrangement order, and dental notation recognition result as a segmentation result.It may include.;

[0142] The step of generating the above segmentation result is a step of performing semantic separation of teeth / gums / background, instance separation of individual tooth objects, distinction between upper and lower jaws, determination of tooth arrangement order, and tooth formula recognition based on the preprocessed first three-dimensional scan data, and then generating the result as structured data that can be input into subsequent steps (e.g., identification of margin lines, generation of prosthetic models).

[0143] Here, the preprocessed first three-dimensional scan data is generally represented in the form of a triangular mesh and may consist of a plurality of vertices and a plurality of faces defined by the connection relationships between the vertices.

[0144] The step of extracting geometric feature vectors is a step of processing local shape information into additional dimension data in addition to simple location information of the original data so that an artificial intelligence model can effectively learn and infer 3D shapes.

[0145] For each vertex and face constituting the preprocessed first 3D scan data, an artificial intelligence model quantifies geometric information meaningful for distinguishing teeth / gums / backgrounds and generates a geometric feature vector.

[0146] Vertices are three-dimensional points that constitute the mesh surface, and faces can be triangular or polygonal patches formed by connecting multiple vertices.

[0147] Geometric feature vectors may include “3D position coordinates (x, y, z)”, “normal vector”, and “curvature” information.

[0148] “3D position coordinates (x, y, z)” represent an absolute position in space and signifies the position occupied by the vertex in the scan coordinate system.

[0149] In order to mitigate the deviation in the value range caused by differences in scanner type or shooting distance, the center of gravity of the entire first three-dimensional scan data may be shifted to the origin and the scale may be defined as a normalized value by dividing the reference length.

[0150] The “normal vector” is the unit direction component that indicates the direction the corresponding vertex or face is facing.

[0151] The normal vector of a surface can be calculated using the cross product of two edge vectors constituting the surface, and the normal vector of a vertex can be calculated by taking an area-weighted average of multiple surface normal vectors adjacent to the vertex. Accordingly, it is possible to distinguish between regions with abrupt shape changes, such as the cusps or incisal edges of teeth, and regions with relatively gentle curves, such as gums.

[0152] “Curvature” is a value representing the degree of curvature of a surface (e.g., pointed cusps of teeth, concave parts of the gum-tooth boundary, etc.). It can be calculated by setting a set of neighboring vertices equal to a pre-set radius or the number of adjacent vertices centered on a specific vertex, fitting a local plane to the distribution of neighboring vertices, and using the degree to which the vertex deviates from the local plane or the directional variance of the neighboring vertices.

[0153] Since curvature information often exhibits characteristics of abrupt change near the boundary between the tooth and the gum, it can contribute to improving the quality of gingival margin and interdental separation.

[0154] The device analyzes the relationship between each vertex of the scan data and its K-Nearest Neighbors to calculate the normal vector and the mean curvature or Gaussian curvature, and combines these with position coordinates to generate a high-dimensional feature vector. Through this, the AI ​​model can clearly distinguish between teeth and gums based solely on geometric features, even in mesh data that lacks color information.

[0155] The step of inputting the above geometric feature vectors into the input layer of the artificial intelligence model is to convert the geometric feature vectors into an input data structure that the artificial intelligence model can process and input them.

[0156] For example, if the number of vertices is very large, a representative set of vertices can be constructed by sampling a portion of the total vertices, and geometric feature vectors can be generated for said representative set of vertices and input. Additionally, to preserve the connectivity of the mesh, connection information in the form of a vertex adjacency list, face adjacency list, or graph can be included as input data.

[0157] Here, the artificial intelligence model can be implemented as any one of a point-based neural network, a graph neural network, or a mesh-based neural network, and the present invention is not limited to a specific architecture.

[0158] The step of calculating the class probability map above is a step of quantifying the probability that each vertex or face belongs to a specific class as a result of the artificial intelligence model's operation and outputting it in the form of a map.

[0159] A “class probability map” is a data structure that holds vector values ​​for each vertex, such as [probability of being a tooth, probability of being a gum, probability of being background]. For example, if a specific vertex has a high probability of being a tooth and a low probability of being a gum, it can be determined that the vertex is highly likely to be classified as a tooth.

[0160] “Background class” includes the tongue, cheeks, lips, or noise caused by the scanner capturing empty space during the scanning process, and by classifying this into a separate class, only the teeth and gums that are the subject of analysis can be precisely extracted.

[0161] The above class probability map can be used for threshold-based filtering, boundary refinement, uncertainty region reprocessing, etc. in subsequent steps, along with the result of simple classification into the class corresponding to the maximum probability value.

[0162] Based on the above class probability map, the step of performing instance segmentation by dividing regions classified as tooth classes into individual tooth objects through connected component analysis and clustering, and assigning instance identifiers to each individual tooth object, is a step of dividing a large mass semantically classified as 'tooth' into individual tooth units (e.g., central incisors, lateral incisors, canines, etc.).

[0163] "Instance segmentation" is a technique that distinguishes different objects even if they belong to the same class (teeth).

[0164] First, to extract vertices or faces classified as tooth classes, elements whose tooth class probability is greater than or equal to a preset tooth threshold can be selected. Subsequently, by performing a connected component analysis using mesh connectivity on the selected elements, interconnected tooth candidate clusters can be primarily extracted.

[0165] However, since connection component analysis alone may result in two teeth merging into a single connection component at the contact site between adjacent teeth, clustering can be additionally performed to separate them into individual tooth objects.

[0166] For example, a method may be applied in which a ridge region with large local curvature (e.g., a deep valley of the tooth boundary) is estimated as a tooth boundary candidate for the tooth candidate region and the region is divided based on said boundary candidate, distance-based clustering is performed on vertices within the tooth candidate region, or the tooth candidate region is overdivided into a number of initial clusters and then the clusters are merged based on tooth shape matching.

[0167] Each separated block is assigned a unique “instance identifier (ID),” enabling analysis and processing at the individual tooth level in subsequent stages.

[0168] Based on the above class probability map, the step of analyzing the curvature of the dental arch and the presence of the palate region to classify whether the first 3D scan data corresponds to the maxilla or the mandible and deriving a result of distinguishing between the upper and lower jaws is a step of determining whether the current data is an upper tooth or a lower tooth as a prerequisite for assigning tooth numbers.

[0169] Even if the AI ​​model does not directly output the 'maxilla / mandibular' class, the device can infer it through geometric rules.

[0170] Specifically, the device can calculate a dental arch curve using the centroid coordinates of individual tooth objects extracted from the instance segmentation results, and analyze the connectivity and space occupancy of the gum class region for the inner region of the dental arch curve.

[0171] In this case, one can utilize the anatomical difference that the maxilla often has a relatively continuous soft tissue surface corresponding to the palate filling the medial side of the dental arch, while the mandible often has an open or hollow medial side of the dental arch that is open toward the tongue space or the floor of the mouth. If the medial area is connected to the gum class, it is determined as 'maxilla (presence of palate)', and if it is empty, as 'mandible (floor of the mouth / tongue space' to derive a result distinguishing the maxilla and mandible.

[0172] However, since exceptions may exist depending on the scan range, the patient's mouth opening status, data omissions, etc., this step can improve the stability of upper and lower jaw differentiation by applying probability-based scoring and correction rules together.

[0173] The step of calculating the centroid coordinates for each of the individual tooth objects and determining the order of tooth arrangement on the entire dental arch using the centroid coordinates is a step of aligning the positional relationships of teeth scattered in three-dimensional space.

[0174] The device calculates the centroid for each tooth by calculating the average of the coordinates of the vertices corresponding to each instance identifier or by taking an area-weighted average of the center points of the faces.

[0175] After that, a curve (dental arch curve) that approximates the distribution of all centers of gravity is generated, and the teeth are aligned in a row according to their position on the curve (e.g., from the left end to the right end) to determine the arrangement order.

[0176] For example, after performing a projection onto a two-dimensional plane on the distribution of centroid coordinates, the dental arch curve can be estimated along the main direction of progression of the projected points, and each centroid coordinate can be aligned by projecting it onto the arc length parameter of the dental arch curve. At this time, the reference for the left and right directions can be determined based on the result of distinguishing between the upper and lower jaws and the geometric center axis of the dental arch curve.

[0177] The step of generating a dental notation recognition result by mapping a unique tooth number to each individual tooth object according to at least one of the FDI (World Dental Federation) notation and the Universal Numbering System, based on the arrangement order on the dental arch and the results of distinguishing between the upper and lower jaws, is a step of assigning a standard number commonly used in dental clinical practice to each tooth.

[0178] The device combines the results of distinguishing between the upper and lower jaws (e.g., upper jaw) and the arrangement order (e.g., 14 teeth present) to map numbers starting from the tooth closest to the midline.

[0179] For example, if it is determined to be the maxilla and is the first tooth to the right of the midline, 'No. 11 (Maxillary Right Central Incisor)' is assigned according to the FDI notation, and 'No. 8' is assigned according to the Universal System. If the result of distinguishing between the maxilla and mandible is determined to be the mandible, a numbering system corresponding to the mandibular dentition may be applied.

[0180] Considering cases where specific teeth are missing due to missing teeth or scan omissions, correction logic can be performed to estimate the location of the missing teeth and reduce errors in tooth mapping by utilizing the abrupt increase in the distance between adjacent centroids, spacing patterns on the dental arch, and shape similarity with a standard tooth library model.

[0181] The step of designating the above class probability map, instance identifier for each individual tooth object, upper and lower jaw distinction result, tooth arrangement order, and tooth formula recognition result as the segmentation result is a step of integrating all information obtained through artificial intelligence analysis into a structured data set and transmitting it to a subsequent step.

[0182] The “segmentation result” goes beyond simple image segmentation and is a set of metadata that includes the meaning of each mesh region (teeth / gums), object identification (instance), affiliation (upper / lower jaw), and name (dental formula).

[0183] For example, the segmentation result can be configured such that class probabilities and class labels are mapped at the vertex or face level, and instance identifiers, centroid coordinates, array order indices, and tooth formula numbers are mapped at the tooth object level.

[0184] Accordingly, in subsequent steps, processing such as margin line identification, separation distance calculation, and prosthetic shape generation can be consistently performed by directly referencing the “tooth object corresponding to a specific tooth number” or the “abutment tooth object.”

[0185] In addition, the step of automatically generating the above 3D design model comprises: a step of obtaining a first standard tooth library model corresponding to the tooth recognition result from a database; a step of generating a second standard tooth library model that is initially placed at the location of the abutment tooth based on the centroid coordinates of the abutment tooth and the curvature of the dental arch, and scaled to correspond to the tooth size of the abutment tooth and the shape of the dental arch; a step of determining the external boundary of the prosthesis by performing mesh deformation on the second standard tooth library model so that the cervical line, which is the lower boundary line of the second standard tooth library model, coincides with the margin line; and a step of generating an intaglio surface mesh offset by a cement gap corresponding to a preset allowable adhesive thickness in the normal direction from the surface mesh of the abutment tooth. The method may include: a step of generating an outer surface mesh by performing shape optimization on the occlusal surface and lateral mesh of the second standard tooth library model to form occlusal points with opposing teeth and contact points with adjacent teeth so as to satisfy spacing distances and design constraints; and a step of combining the outer surface mesh and the inner mesh to generate a three-dimensional design model of a watertight structure having a margin line as a closed curve and an abutment tooth receiving space formed inside.

[0186] The step of automatically generating the above 3D design model is to automatically configure the outer and inner surfaces of a prosthesis that can be mounted on an abutment tooth based on basic design input values ​​such as the margin line and spacing distance, and to output it in the form of a manufacturable 3D mesh model.

[0187] Here, the “3D design model” is mesh data that includes the external and internal shapes of the prosthesis, boundary conditions corresponding to margin lines, and internal reception spaces, and possesses a watertight structure. A watertight structure refers to a state in which the mesh forms a closed surface without openings, clearly distinguishing the interior from the exterior; it is a perfect three-dimensional shape free from holes, self-intersections, or non-manifold structures in the mesh.

[0188] The 3D design model can be implemented, for example, as a triangular mesh, a square mesh, or a mixture of these, and is generated as a file format (STL, OBJ, etc.) that can be output as a physical shape on actual processing equipment (milling machine, 3D printer).

[0189] The step of obtaining a first standard tooth library model corresponding to the above tooth recognition result from the database is a step of retrieving the basic tooth shape that is anatomically most suitable for the patient's abutment tooth location.

[0190] Standard tooth shape data corresponding to the tooth type identified by a specific tooth number in the above tooth recognition result (e.g., incisors, canines, premolars, molars, upper / lower jaw distinction, left / right distinction, etc.) is loaded from the library.

[0191] “The first standard tooth library model” is an original template model selected from a library, which is a set of tooth models with ideal shapes. It can be defined as a three-dimensional template mesh that includes an average crown shape, cusp shape, cervical shape, etc.

[0192] The device uses the tooth formula information recognized in the previous segmentation step as a key value to load a standard model mapped to the corresponding tooth formula from among the numerous libraries stored in the database.

[0193] The standard tooth library model may consist of multiple candidate models, including not only a single average model but also ranges of statistical variation based on gender, age group, racial characteristics, degree of tooth wear, or clinical data. Therefore, logic may be added to select the most similar style library by considering the size or shape of adjacent teeth and the patient's age / gender information.

[0194] The step of initially placing the first standard tooth library model at the location of the abutment tooth based on the center of gravity coordinates of the abutment tooth and the curvature of the dental arch, and generating a second standard tooth library model scaled to correspond to the tooth size of the abutment tooth and the shape of the dental arch, is a step of roughly positioning the loaded standard model to fit the patient's oral environment and adjusting its size.

[0195] The above standard model is aligned with the coordinate system of the patient scan data, and its size and orientation are primarily fitted to the patient's specific shape.

[0196] The initial placement may include translation that moves the reference point of the standard model (e.g., the crown center or cervical center of the standard model) based on the center of gravity coordinates of the abutment tooth.

[0197] The curvature of the dental arch can be used to define the tangential, normal, or center of curvature direction of the dental arch curve, and accordingly, rotational alignment can be performed so that the major axis direction of the standard model (such as the buccolingual axis) is naturally oriented along the dental arch.

[0198] The above scaling can be performed by considering the tooth size of the abutment tooth, the distance from adjacent teeth (mesidistal width), and the distance from opposing teeth (occlusal height). For example, for the mesh area identified as an abutment tooth in the above segmentation results, dimensional indicators such as buccolingual width, mesidistal width, crown height, or distance between specific reference points can be calculated, and a scaling factor can be derived by comparing these with the corresponding dimensional indicators of a standard model.

[0199] In addition to uniform scaling, axial non-uniform scaling can be applied to adjust the scale differently in the long axis direction and the orthogonal direction of the abutment tooth.

[0200] The model that has been initially converted to be patient-customized by reflecting position, rotation, and scale in this way is defined as the “second standard tooth library model.”

[0201] The step of determining the external boundary of the prosthesis by performing mesh deformation on the second standard tooth library model so that the cervical line, which is the lower boundary line of the second standard tooth library model, coincides with the margin line is the step of sewing the bottom edge of the library tooth to fit perfectly with the patient's abutment tooth margin line.

[0202] The boundary conditions of the prosthesis are precisely aligned with the margin line so that the shape is not overextended or underextended at the abutment tooth boundary.

[0203] The bottom of the standard model is a general elliptical shape, but the shape does not match because the patient's margin line is an irregular curve. The “cervical line” can be defined as a closed curve or set of curves near the boundary where the crown and root are separated in the standard model, and can be used as a reference boundary line corresponding to the margin edge of the prosthesis.

[0204] The above mesh deformation may include curve matching that moves each point of the cervical line to a corresponding point of the margin line. For example, the margin line may be sampled at regular intervals to form a margin point sequence, and the cervical line may also be resampled into a sequence of the same or similar number of points, after which a correspondence relationship between the two sequences may be established. The establishment of the correspondence relationship may be implemented by aligning the starting points of the curves, matching the direction of the curve progression, and performing index matching based on the ratio of curve lengths.

[0205] The device forcibly moves the bottom vertices of the second standard tooth library model to corresponding points on the margin line and applies a Laplacian Deformation or Free-form Deformation (FFD) algorithm so that the surrounding meshes naturally follow.

[0206] The above mesh deformation can be performed by combining local and global deformation. For example, a high weight can be applied near the margin line to improve alignment accuracy, while a low weight can be applied to the occlusal surface and the supraclavicular region to prevent excessive distortion of the cusp shape of the standard model.

[0207] To prevent self-intersection or the occurrence of abnormally thin regions during the mesh deformation process, constraints such as the minimum triangle angle, minimum thickness, and rate of change of curvature of the mesh after deformation can be applied.

[0208] This allows the outer boundary of the prosthesis to match the margin of the abutment tooth without gaps, and can improve the marginal fit and the possibility of marginal leakage of the prosthesis.

[0209] The step of creating an intaglio surface mesh offset by a cement gap corresponding to a preset adhesive allowable thickness in the normal direction from the surface mesh of the abutment tooth is a step of creating a cavity for the abutment tooth to enter inside the prosthesis.

[0210] The inner shape is configured to secure a space for adhesive or cement to be received between the inner surface of the prosthesis and the surface of the abutment tooth.

[0211] The “cement gap” can be defined as the target thickness or allowable thickness of the adhesive layer, and can be set, for example, in the range of tens of micrometers to hundreds of micrometers (e.g., 30 to 50 micrometers). This can be set variably depending on the prosthetic material, adhesive type, and clinical protocol.

[0212] The above inner mesh generation can be implemented by generating an offset vertex for each vertex of the abutment tooth surface mesh, which is moved by the cement gap in the normal direction (outward) of the corresponding vertex.

[0213] Since the abutment surface may contain noise or fine irregularities, additional smoothing or curvature-based filtering may be performed on the abutment surface before offset to improve the manufacturability of the inner mesh.

[0214] In particular, since the area near the margin line must be sealed without the adhesive leaking out, a variable cement gap profile (such as Margin Band treatment) may be applied to improve the fit of the marginal area by setting the cement gap small or close to zero, and to improve fit by setting the cement gap relatively large near the occlusal or axial surfaces. In such cases, the cement gap is defined as a value that varies depending on the location and can change stepwise or continuously, for example, according to the distance from the margin line or curvature-based area classification.

[0215] The step of generating an outer surface mesh by performing shape optimization on the occlusal surface and lateral mesh of the second standard tooth library model to form occlusal points with opposing teeth and contact points with adjacent teeth, so as to satisfy spacing distances and design constraints, is a step of functionally refining the external appearance of the prosthesis.

[0216] The shape of the prosthesis's outer surface is adjusted to meet intraoral occlusal and adjacent contact conditions.

[0217] The device deforms the outer mesh using the previously calculated “separation distance” information. Separation distance refers to the minimum gap that must be secured to prevent interference with opposing and adjacent teeth.

[0218] If the occlusal surface of the library model passes through the opposing tooth (interference occurs), recess the area inward to adjust it so that only the occlusal contact is slightly touched. Conversely, if it is too far from the opposing tooth, raise the ridge or cusp to form a proper occlusion.

[0219] The area that must make contact with adjacent teeth (Proximal Contact) is deformed convexly to form a contact point that prevents food from getting trapped.

[0220] The above shape optimization can be performed by configuring the target conditions for the occlusal surface and the lateral surface as different objective functions. For example, in the occlusal surface region, an objective can be set to bring the distance close to a target value in certain areas to form specific occlusal points (contact points), while ensuring that the distance distribution with the opposing tooth surface is greater than or equal to the target occlusal gap. In the lateral surface region, an objective can be set so that the minimum distance with the adjacent tooth surface falls within the target contact distance range (e.g., a range where there is no interference and contact is maintained) to form contact points with adjacent teeth.

[0221] The above shape optimization can be performed using an iterative update method. For example, the nearest point distances and directions to opposing and adjacent teeth can be calculated for the vertices of the outer mesh, and updates can be performed to move vertices where interference occurs inward and vertices where excessive separation occurs outward. In this case, the update size can be adjusted by considering local curvature, Laplacian smoothing terms with adjacent vertices, minimum thickness constraints, etc.

[0222] The most important thing in this process is the “design constraint (minimum thickness).” The device performs shape optimization, but prevents fracture of the prosthesis by imposing a constraint so that the distance between the inner mesh and the outer mesh does not become thinner than a preset minimum thickness threshold (e.g., 0.5 mm).

[0223] Areas below the minimum thickness threshold can be corrected by expanding the outer surface outward or contracting the inner surface inward. However, since inner surface correction may affect the abutment tooth reception space and fitability, in one embodiment, it may be preferable to apply outer surface correction first and perform inner surface correction only in a limited manner when necessary.

[0224] The step of combining the outer mesh and the inner mesh to create a three-dimensional design model of a watertight structure having a margin line as a closed curve and an abutment tooth receiving space formed inside is a step of combining the shell (outer surface) and the inner space (inner surface) into one mass to create a single solid mesh that can be finally manufactured.

[0225] The device creates a single closed mesh by stitching the bottom boundary (margin line) of the outer mesh and the bottom boundary (margin line) of the inner mesh, or by performing a Boolean difference operation between the two meshes.

[0226] The above combination can be implemented by aligning the margin boundary of the inner mesh and the margin boundary of the outer mesh along the margin line, and creating a connecting surface between the two boundaries to form a closed surface. For example, after aligning the outer boundary points and the inner boundary points in equal numbers along the margin line, a connecting mesh in the form of a square strip or a triangular strip connecting adjacent pairs of points can be created.

[0227] After joining, the mesh can be inspected for the presence of fine holes, non-manifold edges, self-intersections, etc., and hole filling, re-triangulation, edge flipping, and local remeshing can be performed to correct the detected defects.

[0228] In the above watertightening process, edge-preserving smoothing or edge-reinforcing treatment may be performed to ensure that the margin line is maintained as an exact closed curve and that the sharpness or chamfer shape of the margin edge is maintained to meet clinical requirements.

[0229] The final 3D design model may further verify the mounting direction of the prosthesis, the presence of undercuts according to the insertion path, and internal curvature limitations considering the minimum tool radius of the manufacturing equipment, and may also include a post-processing loop to fine-tune the outer or inner surface to satisfy design constraints based on the verification results.

[0230] The completed 3D design model is finally transmitted to the machining path generation (CAM) software or rendered on the screen for user verification.

[0231] In addition, the step of deriving the upper and lower jaw distinction result comprises: a step of extracting an area classified as a gum class in the class probability map; a step of calculating a dental arch curve based on the dentition arrangement of the individual tooth objects; a step of establishing an inner region of the dental arch curve based on the dental arch curve and analyzing the connectivity and spatial occupancy of the gum class areas distributed in the inner region; and a step of identifying the inner region as continuous soft tissue constituting the palate and determining the first 3D scan data as the maxilla if, in the connectivity analysis result, the gum class mesh forms a single connectivity component and the spatial occupancy is greater than or equal to a preset occupancy threshold. and if the inner region is classified as a background class, or if the gum class mesh is separated into multiple connection components in the connectivity analysis result, or if the space occupancy is less than the occupancy threshold, the inner region is identified as a space corresponding to the tongue space or floor of mouth, and the first 3D scan data is determined to be the mandible; may be included.

[0232] The step of deriving the result of distinguishing between the upper and lower jaws is a step for automatically determining whether the first three-dimensional scan data corresponds to the upper jaw or the lower jaw, and is a step of performing classification using the curvature of the dental arch and the morphological difference of the soft tissue or space existing inside the dental arch.

[0233] Here, the upper and lower jaws can be distinguished based on the fact that a continuous soft tissue surface corresponding to the palate is observed relatively widely on the medial side of the dental arch in the upper jaw, whereas a spatial area corresponding to the tongue space or the floor of the mouth is likely to be observed on the medial side of the dental arch in the lower jaw.

[0234] Even if an artificial intelligence model has not directly learned the class of 'maxilla' or 'mandible', it is possible to logically infer which part of the oral cavity the scan data belongs to by analyzing the topological layout of the 'teeth' and 'gums'.

[0235] The step of extracting the region classified as the gum class in the above class probability map is to form the gum class region by selecting a vertex or face with a high probability value corresponding to the gum class from the class probability map, which is the output of the artificial intelligence model.

[0236] The process of isolating only the soft tissue region to be analyzed can be performed by, for example, by selecting elements in which the gum class probability is greater than or equal to a pre-set probability threshold (e.g., 0.5).

[0237] Since applying probability thresholds alone may result in residual noise, the gum class region can be refined by removing isolated components of small areas or performing majority-based smoothing of adjacent elements. In this case, the refined gum class region can be defined as a gum class mesh used for subsequent connectivity analysis.

[0238] The step of calculating a dental arch curve based on the dentition arrangement of the individual tooth objects is a step of estimating a curve representing the dental arch using the spatial arrangement of individual tooth objects divided by the instance segmentation.

[0239] It is the process of mathematically defining the overall U-shaped form of the oral cavity. The “dental arch curve” is a curve representing the main path along which the teeth are arranged, and can be approximated, for example, by a second or third-order polynomial curve or a catenary curve model.

[0240] The device generates a dental arch curve by using the centroid coordinates of each tooth object obtained during the instance segmentation phase and fitting the curve that best passes through them.

[0241] This curve serves as a reference line distinguishing the medial (Lingual / Palatal) and lateral (Buccal / Labial) sides of the oral cavity.

[0242] The dental arch curve may be a curve in three-dimensional space, but in one embodiment, considering the planarity of the dental arch, it may be calculated as a two-dimensional curve projected onto an occlusal plane or a reference plane of a scan coordinate system.

[0243] Since specific tooth objects may be omitted or misclassified during the process of calculating the dental arch curve, resulting in the inclusion of outliers, distance deviation detection based on adjacent array order, clustering-based removal, or robust regression-based fitting may be applied to remove outliers.

[0244] The step of establishing an inner region of the dental arch curve based on the above dental arch curve and analyzing the connectivity and spatial occupancy of the gingival class regions distributed in the inner region is a step of defining the “inner” region of the curvature formed by the dental arch curve and quantitatively evaluating whether the gingival class mesh exists continuously within the said region and how widely it occupies it.

[0245] The “Inner Region” is defined as the space inside the dental arch curve, that is, where the tongue or palate is located. The inner region can be established by defining the direction of the center of curvature of the dental arch curve as inner, or by extending the dental arch curve into a closed or semi-closed curve and defining the interior as the inner region.

[0246] For example, an inner boundary curve can be generated by offsetting the dental arch curve by a certain width, and the band area between the dental arch curve and the inner boundary curve can be defined as the inner analysis area. Alternatively, a normal direction vector can be calculated for each sample point of the dental arch curve, and the inner area can be constructed using the trajectory of points moved a predetermined distance in the direction of the normal.

[0247] “Connectivity” is an indicator of whether the gum meshes existing in the inner region are connected as one without being broken. The device considers the gum class mesh as a graph structure to define connection relationships between adjacent faces or vertices, and uses a graph search algorithm (e.g., BFS, DFS) or a Union-Find algorithm to calculate the number of connected components of the gum mesh, the area of ​​each connected component, or the number of vertices.

[0248] “Space occupancy” can be defined as the ratio occupied by the gum class mesh relative to the total area of ​​the inner region or the total number of sample points. For example, the occupancy can be calculated by sampling the inner region with a grid or point cloud of constant resolution and determining whether each sample is included in the gum class mesh.

[0249] In the above connectivity analysis results, if the gum class mesh forms a single connectivity component and the space occupancy is greater than or equal to a preset occupancy threshold, the step of identifying the inner region as continuous soft tissue constituting the palate and determining the first three-dimensional scan data as the maxilla is a step of classification using the anatomical features of the maxilla.

[0250] In the case of the maxilla, the area between the left and right teeth (medial region) is covered by a hard soft tissue called the 'hard palate,' so the gum class is widely distributed without interruption in the scan data.

[0251] Therefore, if the gingival mesh in the medial region is a single large mass (single connecting component) and is densely packed without any gaps (high space occupancy), the device identifies this as the palate and confirms the corresponding data as the maxilla.

[0252] The above occupancy threshold can be set based on statistical values ​​or training data stored in a database, and, for example, different thresholds may be used depending on specific scanning equipment or specific shooting protocols.

[0253] The condition of a single connecting component can be utilized to exclude cases where numerous small segmented gum fragments exist in the medial region, and in one embodiment, robustness may be enhanced by adding an auxiliary condition such as “the occupancy rate of the largest connecting component is greater than or equal to a preset ratio of the total gum occupancy rate.”

[0254] The step of determining the first three-dimensional scan data as the mandible by identifying the inner region as a space corresponding to the tongue space or the floor of the oral cavity when the inner region is classified as a background class, or when the gum class mesh is separated into multiple connection components in the connectivity analysis result, or when the space occupancy is less than the occupancy threshold is a step of classifying using the anatomical features of the mandible.

[0255] In the case of the mandible, the tongue is located between the left and right teeth. During oral scanning, the tongue is often not properly scanned due to movement, resulting in it being treated as a hole (background class) or deleted by the scanner software. Even if it is scanned, the floor of the mouth often takes on a deep and narrow shape, or is discontinuous with the gums surrounding the teeth, rather than being flat and wide like the roof of the mouth.

[0256] The fact that the medial region is classified as a background class may mean that the region was not substantially scanned or was recognized as space, and the fact that the gum class mesh is separated into multiple connected components may mean the possibility that a continuous palatal surface does not exist. Additionally, since a space occupancy rate below a threshold may mean that the area of ​​the gum class in the medial region is not sufficiently secured, it can be determined as the mandible if at least one of these conditions is satisfied.

[0257] Maxilla / Mandibular determination can be implemented not only through a complete binary determination, but also by calculating the probability of being maxillary and mandibular, and then performing the final determination based on the higher probability.

[0258] For example, it is also possible to calculate a classification score by combining multiple feature values ​​such as whether there is a single connection component, occupancy rate, average curvature of the medial region, or height distribution of the medial region (in the case of the maxilla, the possibility of a relative dome shape appearing), and to determine that it is the maxilla if the classification score is above a threshold, and the mandible if it is below the threshold.

[0259] Through this logic, even an AI model that has not been trained on a separate 'palate' dataset can accurately distinguish between the upper and lower jaws using only the 'teeth' and 'gums' classes.

[0260] An apparatus according to one embodiment includes a processor and memory. The processor may include at least one apparatus described above through the drawings or perform at least one method described above through the drawings. The memory may store information related to the method described above or store a program in which the method described above is implemented. The memory may be volatile memory or non-volatile memory.

[0261] The processor can execute a program and control the device. The code of the program executed by the processor can be stored in memory. The device can be connected to an external device (e.g., a personal computer or a network) through an input / output device (not shown in the drawing) and exchange data.

[0262] The embodiments described above may be implemented as hardware components, software components, and / or combinations of hardware and software components. For example, the devices, methods, and components described in the embodiments may be implemented using one or more general-purpose or special-purpose computers, such as, for example, a processor, a controller, an arithmetic logic unit (ALU), a digital signal processor, a microcomputer, a field programmable gate array (FPGA), a programmable logic unit (PLU), a microprocessor, or any other device capable of executing and responding to instructions. The processing unit may execute an operating system (OS) and one or more software applications executed on said operating system. Additionally, the processing unit may access, store, manipulate, process, and generate data in response to the execution of the software. For ease of understanding, the processing unit may be described as being used as a single unit, but those skilled in the art will understand that the processing unit may include multiple processing elements and / or multiple types of processing elements. For example, the processing unit may include multiple processors or one processor and one controller. Additionally, other processing configurations, such as parallel processors, are also possible.

[0263] The method according to the embodiment may be implemented in the form of program instructions that can be executed through various computer means and recorded on a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, etc., either alone or in combination. The program instructions recorded on the medium may be those specifically designed and configured for the embodiment, or they may be those known and available to those skilled in the art of computer software. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes; optical recording media such as CD-ROMs and DVDs; magneto-optical media such as floptical disks; and hardware devices specifically configured to store and execute program instructions, such as ROM, RAM, and flash memory. Examples of program instructions include machine code, such as that generated by a compiler, as well as high-level language code that can be executed by a computer using an interpreter, etc. The hardware devices described above may be configured to operate as one or more software modules to perform the operation of the embodiment, and vice versa.

[0264] Software may include computer programs, code, instructions, or a combination of one or more of these, and may configure a processing unit to operate as desired or instruct the processing unit independently or collectively. Software and / or data may be permanently or temporarily embodied in any type of machine, component, physical device, virtual equipment, computer storage medium, or device so as to be interpreted by the processing unit or to provide instructions or data to the processing unit. Software may be distributed over networked computer systems and stored or executed in a distributed manner. Software and data may be stored on one or more computer-readable recording media.

[0265] Although the embodiments have been described above with reference to the limited drawings, those skilled in the art can apply various technical modifications and variations based on the above. For example, suitable results may be achieved even if the described techniques are performed in a different order than described, and / or if the components of the described system, structure, device, circuit, etc. are combined or assembled in a form different from described, or replaced or substituted by other components or equivalents.

[0266] Therefore, other implementations, other embodiments, and equivalents to the claims also fall within the scope of the claims set forth below. Explanation of the symbols

[0267] 1 : Device (Server) 2 : Database 3: 3D Oral Scanner S100: A step of receiving first 3D scan data and second 3D scan data S200: Step for performing preprocessing S300: Step for generating segmentation results S301: Step to extract geometric feature vectors S302: Step of inputting geometric feature vectors into the input layer of an artificial intelligence model S303: Step for calculating class probability map S304: Step of assigning instance identifiers to individual tooth objects S305: Step for deriving results distinguishing between the upper and lower jaws S306: Step for determining the order of tooth arrangement on the entire dental arch S307: Step for generating dental notation recognition results S308: A step of designating a class probability map, instance identifiers for individual tooth objects, upper and lower jaw distinction results, tooth arrangement order, and tooth formula recognition results as segmentation results. S400: Step to identify the margin line S500: Step for calculating separation distance S501: Step to identify the mesh area S502: A step of extracting the surface mesh of the antagonist tooth and the surface mesh of the adjacent tooth adjacent to the abutment tooth, respectively. S503: A step of aligning the abutment tooth and the opposing tooth so that they are placed on the same coordinate system. S504: Step of determining the subquantile values ​​of projection distances as the occlusal clearance S505: Step for calculating the occlusal plane S506: Step to generate horizontal normal vector S507: Step of determining the lower quantile value of the lateral distances as the Proximal Clearance S508: Step of determining the minimum value between the occlusal gap and the adjacent gap as the separation distance S509: Step to obtain minimum thickness threshold S510: Step of setting 'prosthesis minimum thickness ≥ minimum thickness threshold' as a design constraint S511: Step to output material candidate information or generate tooth removal guide information S600: Step to automatically generate a 3D design model S601: Step of obtaining the first standard tooth library model from the database S602: Step to generate a second standard tooth library model S603: Step for determining the external boundaries of the prosthesis S604: Step to generate the Intaglio Surface Mesh S605: Step to generate the outer surface mesh S606: Step to create a 3D design model

Claims

Claim 1 A method for providing an automated solution for tooth analysis and modeling based on 3D scanning data utilizing an artificial intelligence model, wherein a device includes a processor, memory, a communication module, and a non-transient storage medium, and the program stored in the non-transient storage medium is executed by the processor, comprising: a step of receiving first 3D scan data generated by scanning an oral region including a prepared tooth, which is the target of prosthetic fabrication, in 3D, and second 3D scan data generated by scanning an oral region including an antagonist tooth in 3D; a step of generating a segmentation result by processing the distinction between the maxilla and mandible, the separation of the tooth region and the gum region, and the recognition of dental notation corresponding to each tooth from the first 3D scan data using an artificial intelligence model; a step of identifying a margin line, which is the boundary between the prepared tooth and the gum, which is the target of prosthetic fabrication, based on the segmentation result; and the first 3D scan data and the second 3D The method comprises: a step of registering scan data to determine whether the abutment tooth interferes with the opposing tooth and calculating a separation distance secured for the formation of a prosthesis; and a step of automatically generating a three-dimensional design model of a dental prosthesis to be mounted on the abutment tooth based on the margin line and the separation distance; wherein the step of calculating the separation distance includes: a step of identifying mesh regions constituting the occlusal surface, mesial surface, and distal surface of the abutment tooth, respectively, based on the segmentation results; and a step of extracting the surface mesh of the antagonist tooth and the surface mesh of the adjacent tooth adjacent to the abutment tooth, respectively.A step of aligning the first 3D scan data and the second 3D scan data so that the abutment tooth and the opposing tooth are placed on the same coordinate system; a step of measuring the projection distance from each vertex of the mesh constituting the occlusal surface of the abutment tooth to the surface of the opposing tooth along the normal direction of each vertex of the occlusal surface, and determining the lower quantile value of the projection distances as the occlusal clearance; a step of calculating an occlusal plane by performing least squares plane fitting on the set of vertices of the mesh constituting the occlusal surface of the abutment tooth; and a step of generating a horizontal normal vector by projecting the normal vector of the set of lateral vertices, which are the vertices of the mesial and distal surfaces of the abutment tooth, onto the occlusal plane. A method for providing an automated tooth analysis and modeling solution based on 3D scanning data utilizing an artificial intelligence model, comprising: a step of generating a ray in the direction of the horizontal normal vector to search for the nearest intersection point between the ray and the surface mesh of an adjacent tooth, calculating lateral distances which are the distances between each element of a lateral vertex set and the nearest intersection point, and determining the lower quantile value of the lateral distances as the proximal clearance; a step of determining the minimum value among the occlusal clearance and the proximal clearance as the separation distance; a step of obtaining a preset minimum thickness threshold based on material information of the prosthesis to be fabricated; a step of determining, if the separation distance is greater than or equal to the minimum thickness threshold, that the state of being able to design a prosthesis based on the material is determined, and setting 'prosthesis minimum thickness ≥ minimum thickness threshold' as a design constraint for generating a 3D design model; and a step of determining, if the separation distance is less than the minimum thickness threshold, that the state of being unable to design a prosthesis based on the material is determined, and generating tooth reduction guide information including occlusal surface reduction amount, mesial surface reduction amount, and distal surface reduction amount. Claim 2 In claim 1, the step of generating the segmentation result comprises: extracting a geometric feature vector including 3D position coordinates (x, y, z), a normal vector, and curvature information for each of a plurality of vertices and faces constituting the preprocessed first 3D scan data; inputting the geometric feature vector into an input layer of an artificial intelligence model; calculating a class probability map through the operation of the artificial intelligence model, which indicates the probability that each vertex and face belongs to one of the Tooth Class, Gingiva Class, and Background Class; and, based on the class probability map, performing instance segmentation by dividing the regions classified as tooth classes into individual tooth objects through connected component analysis and clustering, and assigning an instance identifier for each individual tooth object. A step of analyzing the curvature of the dental arch and the presence of the palate region based on the class probability map above to classify whether the first 3D scan data corresponds to the maxilla or the mandible, thereby deriving a result of distinguishing between the upper and lower jaws; a step of calculating the centroid coordinates for each of the individual tooth objects and determining the tooth arrangement order on the entire dental arch using the centroid coordinates; and a step of generating a dental notation recognition result by mapping a unique tooth number to each individual tooth object according to at least one of the FDI (World Dental Federation) notation and the Universal Numbering System, based on the arrangement order on the dental arch and the result of distinguishing between the upper and lower jaws.A method for providing an automated tooth analysis and modeling solution based on 3D scanning data utilizing an artificial intelligence model, comprising the step of designating the class probability map, instance identifier for each individual tooth object, upper and lower jaw distinction result, tooth arrangement order, and tooth formula recognition result as a segmentation result; Claim 3 delete

Images