JP7299067B2 - DATA GENERATION DEVICE, SCANNER SYSTEM, DATA GENERATION METHOD, AND DATA GENERATION PROGRAM - Google Patents

Description

The present invention relates to a data generation device, a scanner system , a data generation method, and a data generation program.

2. Description of the Related Art Conventionally, in the field of dentistry, a three-dimensional scanner for acquiring a three-dimensional shape of a site including a missing tooth is known in order to digitally design a prosthesis or the like on a computer. For example, Patent Literature 1 discloses a technique of recording the shape of the oral cavity by imaging the intraoral cavity using a three-dimensional camera of a three-dimensional scanner.

JP-A-2000-74635

An operator such as a dentist can record shape data (hereinafter also referred to as "three-dimensional data") of a patient's oral cavity by using the technology disclosed in Patent Document 1. The recorded three-dimensional data is used to fabricate a prosthesis for filling a missing tooth or the like. For example, the operator acquires three-dimensional data including the patient's missing tooth with a three-dimensional scanner, and based on the acquired three-dimensional data, creates data (hereinafter referred to as " The prosthesis data") is digitally designed on a computer. Alternatively, the operator sends the acquired three-dimensional data to a dental technician, and the dental technician digitally designs the prosthesis data on a computer to create a prosthesis that fits the defect site based on the three-dimensional data. do.

However, since the skill levels of operators and dental technicians vary, it is not always easy to generate appropriate prosthetic data for the defect site based on the acquired three-dimensional data. A technology that can create objects is desired.

The present invention has been made to solve such problems, and provides a data generating apparatus, a scanner system , a data generating method, and a data generating method that can produce an appropriate prosthesis more easily. The purpose is to provide a generation program.

According to one example of the present disclosure, a data generation device is provided for generating three-dimensional data for fabricating a dental prosthesis . The data generation device scans the adjacent teeth for each of a plurality of teeth in which the missing tooth, which is a missing tooth, and peripheral teeth that are located around the missing tooth and do not include the missing tooth are adjacent to each other. An acquisition unit that acquires three-dimensional data including three-dimensional position information for each of a plurality of points that constitute the scan target region for indicating the three-dimensional shape of the scan target region, and extracts latent variables from the plurality of data. Extraction that extracts latent variables related to the characteristics of peripheral teeth for each of a plurality of adjacent teeth from the three- dimensional data acquired by the acquisition unit based on an extraction model that includes the first algorithm in VAE where machine learning is performed for an accumulator for preliminarily accumulating latent variables relating to tooth characteristics for each of a plurality of adjacent teeth; and a second algorithm in VAE in which machine learning is performed to generate data based on the plurality of latent variables. Based on the model, the latent variables related to the peripheral tooth features of each of the adjacent teeth extracted by the extraction unit and the latent variables related to the tooth features of each of the adjacent teeth accumulated by the storage unit are used. and a generator for generating three-dimensional data for producing a prosthesis that fits the missing part of the missing tooth .

According to one example of the present disclosure, a scanner system is provided for generating three-dimensional data for fabricating a dental prosthesis . The scanner system uses a three-dimensional camera to scan for each of a plurality of adjacent teeth a missing tooth , which is a missing tooth, and peripheral teeth that are located around the missing tooth and do not include the missing tooth. a three-dimensional scanner for acquiring three-dimensional data including three-dimensional position information at each of a plurality of points constituting the scan target region for indicating the three-dimensional shape of the scan target region scanned in the tooth; and a data generating device for generating three -dimensional data for manufacturing a prosthesis that fits the missing part of the missing tooth based on the three-dimensional data acquired by the scanner. The data generation device includes an acquisition unit that acquires three-dimensional data including three-dimensional position information at each of a plurality of points that constitute a scan target region acquired by a three-dimensional scanner, and a latent variable that extracts latent variables from the plurality of data. Extraction that extracts latent variables related to the characteristics of peripheral teeth for each of a plurality of adjacent teeth from the three- dimensional data acquired by the acquisition unit based on an extraction model that includes the first algorithm in VAE where machine learning is performed for an accumulator for preliminarily accumulating latent variables relating to tooth characteristics for each of a plurality of adjacent teeth; and a second algorithm in VAE in which machine learning is performed to generate data based on the plurality of latent variables. Based on the model, the latent variables related to the peripheral tooth features of each of the adjacent teeth extracted by the extraction unit and the latent variables related to the tooth features of each of the adjacent teeth accumulated by the storage unit are used. and a generator for generating three-dimensional data for fabricating the prosthesis .

According to one example of the present disclosure, a data generation method is provided for computer generating three-dimensional data for fabricating a dental prosthesis. In the data generation method, as a process executed by a computer, for each of a plurality of teeth in which a missing tooth that is a missing tooth and peripheral teeth that are located around the missing tooth and do not include the missing tooth are adjacent, the adjacent tooth a step of acquiring three-dimensional data including three-dimensional positional information of each of a plurality of points constituting the scan target region for indicating the three-dimensional shape of the scan target region scanned in the plurality of teeth; From the three-dimensional data acquired by the step of acquiring , based on an extraction model comprising a first algorithm in VAE where machine learning was performed to extract latent variables from adjacent multiple and based on a generative model including a second algorithm in VAE where machine learning is performed to generate data based on a plurality of latent variables. To create a prosthesis that fits the missing part of the missing tooth by using the latent variables related to the characteristics of the peripheral teeth of each tooth and the latent variables related to the characteristics of the teeth of multiple adjacent teeth that have been accumulated in advance . and generating three-dimensional data of

According to one example of the present disclosure, a data generation program is provided for generating three-dimensional data for fabricating a dental prosthesis. A program for generating data causes a computer to generate a plurality of adjacent teeth for each of a plurality of teeth adjacent to a missing tooth that is a missing tooth and peripheral teeth that are located around the missing tooth and do not include the missing tooth. acquiring three-dimensional data including three-dimensional position information at each of a plurality of points constituting the scan target region for indicating the three-dimensional shape of the scan target region scanned in; From the three-dimensional data acquired by the acquiring step, the latent variables related to the characteristics of the peripheral teeth are extracted from the three- dimensional data acquired by the acquiring step based on the extraction model including the first algorithm in VAE where machine learning is performed to extract the and a plurality of adjacent teeth extracted by the step of extracting based on a generative model comprising a second algorithm in a machine-learned VAE to generate data based on a plurality of latent variables Three-dimensional prosthesis for creating a prosthesis that fits the missing part of the missing tooth and generating data.

According to the present invention, a suitable prosthesis can be produced more easily.

1 is a schematic diagram showing an application example of a data processing device according to an embodiment; FIG. FIG. 2 is a schematic diagram showing an example of a tooth to be scanned by a three-dimensional scanner according to the present embodiment; FIG. 4 is a schematic diagram for explaining types of teeth to be scanned by the three-dimensional scanner according to the present embodiment; 3 is a schematic diagram showing a functional configuration in a learning stage of the data processing device according to the present embodiment; FIG. FIG. 3 is a schematic diagram showing the functional configuration of a data generator according to the embodiment; 1 is a schematic diagram showing a functional configuration of a data processing device according to an embodiment in a practical stage; FIG. 1 is a schematic diagram showing the overall configuration of a system according to an embodiment; FIG. 1 is a schematic diagram showing a hardware configuration of a data processing device according to an embodiment; FIG. FIG. 4 is a schematic diagram for explaining accumulation of latent variable data; FIG. 4 is a schematic diagram for explaining accumulation of latent variable data; FIG. 4 is a schematic diagram showing an example of accumulated latent variable data; FIG. 4 is a schematic diagram showing an example of prosthesis data generation in the learning stage; FIG. 4 is a schematic diagram showing an example of prosthesis data generation in a practical stage; 4 is a flowchart for explaining learning processing executed by the data processing device according to the embodiment; 4 is a flowchart for explaining data generation processing executed by the data processing device according to the embodiment;

Embodiments of the present invention will be described in detail with reference to the drawings. The same or corresponding parts in the drawings are given the same reference numerals, and the description thereof will not be repeated.

[Application example]
FIG. 1 is a schematic diagram showing an application example of a data processing apparatus according to this embodiment.

As shown in FIG. 1 , the user 1 can use the scanner system 10 to obtain three-dimensional shape data (three-dimensional data) of the oral cavity including the teeth of the subject 2 . It should be noted that the "user" is any person who uses the scanner system 10, such as an operator such as a dentist, a dental assistant, a teacher or student at a dental college, a dental technician, a manufacturer's technician, or a factory worker. may be The “subject” may be any person to whom the scanner system 10 is applied, such as a patient at a dental clinic or a subject at a dental college. Also, the "subject" is not limited to a real person, and may be a human body model or a head model.

Scanner system 10 according to the present embodiment includes three-dimensional scanner 200 , data processing device 100 , and display 300 . The three-dimensional scanner 200 acquires three-dimensional data of a scanning target by a built-in three-dimensional camera. Specifically, the three-dimensional scanner 200 scans the oral cavity of the subject 2 to obtain positional information (for example, lateral direction, depth direction, height direction) and color information (for example, the color of the tooth surface) are acquired using an optical sensor or the like. The data processing device 100 generates a 3D image based on the 3D data acquired by the 3D scanner 200 and displays the generated 3D image on the display 300 .

Specifically, the user 1 uses a three-dimensional scanner 200 to digitally design prosthesis data on a computer for fabricating a prosthesis that fits the missing part of the missing tooth of the subject 2. Imaging the inside of the oral cavity. Each time the user 1 takes an image of the oral cavity, three-dimensional data is sequentially obtained, and a three-dimensional image of the oral cavity including at least the missing tooth is displayed on the display 300 . The user 1 confirms the three-dimensional image displayed on the display 300 and scans the missing part of the three-dimensional data. The three-dimensional data recorded in this way are used for manufacturing the prosthesis.

A "missing tooth" includes, for example, an abutment tooth, a cavity, and an implant body, whose shape has been lost due to caries or cutting during treatment. "Prostheses" include, for example, crowns, bridges, implants, inlays, and other known various fillings and caps employed in prosthetic dentistry.

For example, when prostheticing a missing tooth with a crown, an operator first forms an abutment tooth by scraping off the carious portion of the tooth with a cutting instrument such as a turbine, and then mounts a crown thereon. At this time, if the outer edge of the abutment tooth, called the margin, and the outer edge of the crown are not brought into close contact with each other, secondary caries may occur. Similarly, in the case of using an inlay to prosthetic a defect, secondary caries may occur if the inlay is not brought into close contact with the cavity after the carious portion of the tooth has been scraped away.

In addition, in the production of a prosthesis, the teeth around the missing part (missing tooth) (hereinafter also referred to as "peripheral teeth"), that is, the teeth adjacent to the missing tooth (hereinafter also referred to as "adjacent teeth") and the prosthesis It is also important that the distance between objects and the distance between the tooth opposite the missing part (missing tooth) (hereinafter also referred to as "opposing tooth") and the prosthesis is appropriate. Note that "adjacent" includes being positioned next to the missing tooth while being in contact with the missing tooth, or being positioned next to the missing tooth without being in contact with the missing tooth.

Furthermore, although details will be described later, teeth include central incisors, lateral incisors, canines, first premolars, second premolars, first molars, second molars, and third molars. , there are various types. In addition, the shape of teeth is characteristic according to the type of tooth, and these shapes affect the meshing of the teeth. Therefore, it is important to produce a prosthesis of the same type as the missing tooth, that is, a prosthesis having the same shape (characteristics) as the missing tooth as much as possible.

As described above, there are many important points in the production of a prosthesis, but since the skill level of the operator and the dental technician who are the user 1 varies, it is necessary to generate prosthesis data appropriate for the defect site. It's not always easy.

Therefore, the scanner system 10 according to the present embodiment utilizes AI (Artificial Intelligence) of the data processing device 100, based on the three-dimensional data acquired by the three-dimensional scanner 200, appropriate for the missing portion. prosthesis data.

Specifically, when the user 1 scans the oral cavity of the subject 2 using the three-dimensional scanner 200 , three-dimensional data including at least the missing tooth is input to the data processing device 100 . The data processing device 100 automatically generates prosthesis data that fits the defect site based on the input three-dimensional data and generation model. The prosthesis data, as three-dimensional data for fabricating the prosthesis itself, includes positional information (coordinates of each axis in the longitudinal, lateral, and height directions) of each of a plurality of points forming the prosthesis. A PCD file, an STL file, a PLY file, or the like is applied as the output format of the prosthesis data.

The prosthesis data generated by the data processing device 100 in this manner is output to the dental laboratory. At the dental laboratory, a dental technician fabricates a prosthesis based on the prosthesis data acquired from the data processing apparatus 100 .

Further, when an automatic manufacturing device 600 capable of automatically manufacturing a prosthesis is arranged in the dental clinic, the prosthesis data generated by the data processing device 100 may be output to the automatic manufacturing device 600. In this way, the user 1 can manufacture the prosthesis based on the prosthesis data also by the automatic manufacturing device 600 . Examples of automated manufacturing equipment 600 include milling machines and 3D printers.

As described above, according to the scanner system 10 according to the present embodiment, the AI of the data processing device 100 is used to automatically generate prosthesis data based on the three-dimensional data acquired by the three-dimensional scanner 200. be done. By using AI, the scanner system 10 can also find tooth features that cannot be extracted by the user 1, so that the user 1 can more easily obtain an appropriate prosthesis.

[An example of a tooth to be scanned]
FIG. 2 is a schematic diagram showing an example of a tooth to be scanned by the three-dimensional scanner 200 according to this embodiment. In FIG. 2, the tooth to be scanned by the three-dimensional scanner 200 is represented by a diagram.

As shown in FIG. 2, when the tooth to be scanned by the three-dimensional scanner 200 is the maxillary incisor, the three-dimensional images obtained are images of the upper lip side, the palate side, and the incisal side. The user 1 scans the oral cavity of the subject 2 so as to include at least Also, when the teeth to be scanned by the three-dimensional scanner 200 are the canines and molars of the upper jaw, the obtained three-dimensional image should include at least images of the buccal region, the palatal region, and the occlusal region. , the oral cavity of a subject 2 is scanned by a user 1 . When the tooth to be scanned by the three-dimensional scanner 200 is a mandibular incisor, the three-dimensional image obtained includes at least images of the lower lip side region, the lingual region, and the incisal side region. A user 1 scans the oral cavity of a subject 2 . When the teeth to be scanned by the three-dimensional scanner 200 are the canines and molars of the mandible, the resulting three-dimensional image includes at least images of the buccal region, the lingual region, and the occlusal region. 1 scans the oral cavity of a subject 2 .

In general, the teeth of the subject 2 differ in shape and size depending on their type. For example, for maxillary incisors, the upper labial surface is generally U-shaped, whereas for maxillary canines, the buccal surface is generally pentagonal. Each tooth is characteristic in shape and size according to its type. Therefore, digitally designing prosthesis data while ascertaining the characteristics (types) of each tooth greatly depends on the knowledge of the operator and the dental technician.

[Type of tooth to be scanned]
FIG. 3 is a schematic diagram for explaining the types of teeth to be scanned by the three-dimensional scanner 200 according to this embodiment.

As shown in FIG. 3, each tooth has a central incisor, a lateral incisor, a canine, a first premolar, a second premolar, a first molar, a second molar, and third molars. Further, in the oral cavity, the above-described teeth are generally present on each of the right upper jaw, the left upper jaw, the right lower jaw, and the left lower jaw.

Furthermore, each tooth is assigned a predetermined number according to its type and position. For example, the central incisor is assigned number 1, the lateral incisor is assigned number 2, the canine is assigned number 3, the first premolar is assigned number 4, the second premolar is assigned The number 5 is assigned, the first molar is assigned the number 6, the second molar is assigned the number 7, and the third molar is assigned the number 8.

[Functional Configuration in Learning Stage of Data Processing Device]
FIG. 4 is a schematic diagram showing the functional configuration in the learning stage of data processing apparatus 100 according to the present embodiment. The data processing device 100 may have a function of generating prosthesis data for manufacturing a tooth prosthesis, and is also called a "data generation device".

As shown in FIG. 4 , the data processing device 100 has an input section 1102 , a data generation section 1104 and an identification section 1106 . These functions are realized by the arithmetic device 130 of the data processing device 100, which will be described later, executing the OS 127, the identification program 120, the learning program 121, the data processing program 125, and the like. The data processing program 125 may include processing for generating prosthesis data, and is also called a "data generation program".

Three-dimensional data of teeth is input to the input unit 1102 . The three-dimensional data input to the input unit 1102 include an arbitrary tooth, an adjacent tooth adjacent to the arbitrary tooth, an opposing tooth opposite to the arbitrary tooth, a tooth adjacent to the opposite tooth, and an opposite tooth to the arbitrary tooth. 3D data of at least one of the side teeth is included.

For example, referring to FIG. 3, if an arbitrary tooth is the maxillary left canine (number 3), each tooth can be exemplified as follows. For example, the adjacent teeth adjacent to the maxillary left canine (3rd) include the maxillary left lateral incisor (2nd), central incisor (1st), first premolar (4th), or second minor tooth. Molar tooth (number 5) and the like. The opposing teeth facing the upper left canine (3rd) include the lower left canine (3rd). Examples of the teeth adjacent to the opposing teeth facing the maxillary left canine (number 3) include the mandibular left lateral incisor (number 2) or the first premolar (number 4). Further, the tooth on the opposite side of an arbitrary tooth is the tooth on the left and right opposite sides in the row of teeth to which the arbitrary tooth belongs. For example, the tooth on the opposite side of the maxillary left canine (number 3) includes the maxillary right canine (number 3).

Note that the input unit 1102 is not limited to the actual three-dimensional data of the tooth acquired by the three-dimensional scanner 200, and a three-dimensional data model prepared in advance for learning may be input. Furthermore, the input unit 1102 stores the type (name or number) (for example, tooth name or number) is input.

As shown in FIG. 4, as an example of the learning stage, the input unit 1102 stores three-dimensional data of each tooth in a tooth row in which an arbitrary tooth is omitted, and the type (name or number) of each tooth in the tooth row. Data may be entered to identify the

The data generator 1104 includes a generator 1144 that generates three-dimensional data of teeth. For example, the generation unit 1144 generates three-dimensional data (hereinafter also referred to as prosthesis data) suitable for any tooth based on the three-dimensional data input from the input unit 1102 and the generation model 1144a. Such processing for generating prosthesis data by the generation unit 1144 is also referred to as “data generation processing”. When the type (name or number) of each tooth included in the three-dimensional data is input to the input unit 1102, the generation unit 1144 generates the type (name or number) of each tooth included in the three-dimensional data input by the user. or number) to generate prosthesis data.

Here, the data generator 1104 will be specifically described with reference to FIG. FIG. 5 is a schematic diagram showing the functional configuration of data generating section 1104 according to this embodiment. As shown in FIG. 5, the data generation unit 1104 has an acquisition unit 1141 , an extraction unit 1142 , an accumulation unit 1143 and a generation unit 1144 .

Acquisition unit 1141 acquires three-dimensional data input from input unit 1102 . The extraction unit 1142 extracts latent variables related to tooth features from the three-dimensional data acquired by the acquisition unit 1141 based on the extraction model 1142a.

Specifically, the extraction model 1142a includes a machine-learned algorithm (corresponding to the “first algorithm”) for extracting latent variables from a plurality of data. For example, the extraction model 1142a includes a neural network (not shown) and parameters used by the neural network. In the neural network of the extraction model 1142a, processing by deep learning is performed by the intermediate layer having a multi-layered structure. Note that an existing neural network mechanism of the extraction model 1142a may be applied.

The extraction unit 1142 has a function called an encoder, and analyzes the three-dimensional data of the tooth acquired by the acquisition unit 1141 based on the extraction model 1142a. Based on the results, latent variables that characterize the type of tooth are extracted as latent variables from the three-dimensional tooth data. For example, the extraction unit 1142 extracts information such as tooth shape (overall size, overall shape, edge shape, etc.), tooth position, tooth color, relationship with surrounding teeth (positional relationship, degree of contact, etc.). For each characterizing element, the characteristic is extracted as a latent variable. Note that the extracting unit 1142 is not limited to extracting one latent variable each time three-dimensional data is acquired, and may extract a plurality of latent variables each time three-dimensional data is acquired. good too.

The accumulation unit 1143 accumulates data for identifying the latent variables extracted by the extraction unit 1142 (hereinafter also referred to as “latent variable data”). Note that accumulation of latent variable data in the accumulation unit 1143 will be described later with reference to FIGS. 9 and 10. FIG.

The generative model 1144a of the generator 1144 includes an algorithm (corresponding to the “second algorithm”) subjected to machine learning to generate data based on multiple latent variables. For example, generative model 1144a includes a neural network (not shown) and parameters used by the neural network. In the neural network of the generative model 1144a, processing by deep learning is performed because the intermediate layer has a multi-layered structure. Note that an existing neural network mechanism of the generative model 1144a may be applied.

The generation unit 1144 has a function called a decoder, and when the acquisition unit 1141 acquires three-dimensional data of each tooth in a tooth row from which an arbitrary tooth is omitted, the generation unit 1144 extracts an arbitrary tooth based on the generation model 1144a Peripheral teeth (e.g. adjacent teeth adjacent to any tooth, opposing teeth opposing any tooth, teeth adjacent to opposing teeth, teeth opposite to any tooth) Using the latent variables and the latent variables specified by the latent variable data already accumulated by the accumulation unit 1143, three-dimensional data (prosthesis data) suitable for any tooth is generated. For example, the generation unit 1144 generates three-dimensional data (prosthesis data) of teeth of the same type (name or number) as the type (name or number) of an arbitrary tooth.

In the data generation unit 1104 described above, programs for performing processing specialized for three-dimensional images, such as AAE, “Learning Representations and Generative Models”, ShapeVAE, “Shape Generation using Spatially Partitioned Point Clouds”, FoldingNet, P2P- Net, PCN (Point Completion Network), PPF-FoldingNet, PC-GAN, and DeepSDF are used, but other programs may be used.

Returning to FIG. 4 , the prosthesis data generated by the generation unit 1144 is output to the identification unit 1106 . The identification unit 1106 identifies whether or not the prosthesis data is appropriate based on the prosthesis data generated by the generation unit 1144 and the correct data input by the user. To identify whether the prosthesis data is appropriate or not, it is necessary to identify whether the shape of the prosthesis manufactured based on the prosthesis data matches the shape of the sample tooth input as the correct data; To discriminate whether or not the degree of similarity between the shape of a prosthesis manufactured based on prosthesis data and the shape of a sample tooth input as correct data is equal to or greater than a reference value; identifying whether or not the tooth type (name or number) of the prosthesis to be used matches any tooth type (name or number) input as correct data. Such a process of identifying whether or not the prosthesis data is appropriate by the identification unit 1106 is also called "identification process".

In this embodiment, the identification unit 1106 determines whether the shape of the prosthesis to be manufactured based on the prosthesis data generated by the generation unit 1144 matches the shape of the sample tooth input as the correct data. identify

For example, if the number 6 first molar on the right side of the mandible is omitted as an arbitrary tooth, the user 1 sends three-dimensional data corresponding to the number 6 first molar to the data processing device 100 as correct data. Enter in advance. In the data processing device 100, when the three-dimensional data of the oral cavity in which the sixth first molar is omitted is input, the generation unit 1144, based on the input three-dimensional data and the generation model 1144a, An attempt is made to generate prosthesis data for making a prosthesis with the characteristics of the number 6 first molars as much as possible. The identification unit 1106 compares the prosthesis data generated by the generation unit 1144 with the three-dimensional data corresponding to the first molar No. 6 previously input by the user, and identifies whether the two match. do.

The identification result obtained by identification section 1106 is fed back to generation section 1144 . The generative model 1144a is machine-learned based on the identification results fed back from the identification unit 1106. FIG. Such a process of performing machine learning on the generative model 1144a is also referred to as a "learning process".

For example, in the learning process, the generative model 1144a does not update the parameters if it is determined that the prosthesis data generated by the generating unit 1144 matches the correct data based on the identification result. The parameters are optimized by updating the parameters so that Note that the generative model 1144a is not limited to the case where the prosthetic data generated by the generating unit 1144 and the correct data completely match, and the degree of matching between the prosthetic data generated by the generating unit 1144 and the correct data is predetermined. It may be determined that the two match when the reference value of is exceeded. Thus, in the learning process, the generative model 1144a is machine-learned by optimizing the parameters. Note that in the generative model 1144a, parameters are not limited to being updated, and a neural network (for example, a neural network algorithm) may be updated.

In the data processing apparatus 100 configured as described above, three-dimensional data acquired by the three-dimensional scanner 200 or three-dimensional data prepared in advance for learning is input. Before learning, the generation unit 1144 cannot generate appropriate prosthesis data that fits any omitted tooth, but first, according to its own prediction, based on the three-dimensional data and the generation model 1144a. Try to generate some data. The identification unit 1106 identifies whether or not the prosthesis data generated by the generation unit 1144 matches the correct data, and feeds back the identification result to the generation unit 1144 . The generation unit 1144 machine-learns the generative model 1144a based on the feedback identification result, thereby generating appropriate prosthesis data closer to the correct data than before learning. In addition, by repeating such machine learning, the accuracy of the generation of prosthesis data by the generation unit 1144 is improved.

Furthermore, when the type (name or number) of each tooth included in the three-dimensional data, such as an arbitrary tooth, an adjacent tooth, an opposing tooth, an opposing tooth and adjacent teeth, and an opposing tooth, is input to the input unit 1102 , the generation unit 1144 generates prosthesis data based on the type (name or number) of each tooth included in the three-dimensional data. In this case, the generation unit 1144 performs machine learning in consideration of the type (name or number) of each tooth input by the user, thereby generating more accurate prosthesis data.

[Functional Configuration of Data Processing Device in Practical Stage]
FIG. 6 is a schematic diagram showing the functional configuration of the data processing device 100 according to this embodiment in the practical stage.

Each time the generative model 1144a in the generator 1144 undergoes machine learning through the learning process described with reference to FIGS. Become. When the generation unit 1144 can generate prosthesis data for making an appropriate prosthesis that meets the criteria, the user will use the data processing device 100 in the practical stage. In other words, when the scanner system 10 equipped with the trained data processing device 100 is put into the market, the prosthesis data for fabricating the prosthesis of the actual patient is acquired by the operator who is the user 1 .

As shown in FIG. 6, since machine learning of the generative model 1144a is not required in the practical stage, the data processing device 100 does not need to include the identification unit 1106 involved in the learning process. When the operator uses the three-dimensional scanner 200 to acquire three-dimensional data of the actual patient's oral cavity, the acquired three-dimensional data is input from the input unit 1102 . The generation unit 1144 generates prosthesis data based on the three-dimensional data input from the input unit 1102 and the learned generation model 1144a. The prosthesis data generated by the generator 1144 is sent to the dental laboratory or the automatic manufacturing device 600 . Then, based on the prosthesis data, a prosthesis suitable for the defect in the oral cavity of the patient is manufactured. After the prosthesis data is generated, when the user fine-tunes the generated prosthesis data to finally complete the prosthesis, the generated model 1144a is machine-learned using the completed prosthesis data as correct data. may

Note that the data processing apparatus 100 may include the identification unit 1106 even in the practical stage, and may execute the learning process in the practical stage as well. In this way, even after the scanner system 10 equipped with the data processing device 100 is put on the market as a product, the generative model 1144a is machine-learned based on the three-dimensional intraoral data of the actual patient. be able to. As a result, the generation accuracy of prosthesis data by the generation unit 1144 can be improved in the practical stage.

[Overall system configuration]
FIG. 7 is a schematic diagram showing the overall configuration of the system according to this embodiment.

As shown in FIG. 7, the scanner system 10 is arranged at each of a plurality of locals A-C. For example, local A and local B are dental clinics, and in the dental clinics, an operator who is user 1 acquires three-dimensional data including teeth of a patient who is subject 2 using scanner system 10. At the same time, prosthesis data is generated. Local C is a dental college, and at the dental college, a teacher or a student who is user 1 acquires intraoral three-dimensional data of a subject who is subject 2 and generates prosthesis data. The intraoral three-dimensional data and prosthesis data acquired by each of the locals A to C are transmitted to the dental laboratory, which is the local D, via the network 5 .

The intraoral three-dimensional data and the prosthesis data acquired by each of the locals A to C may be output via the network 5 to the server device 500 arranged in the management center. In the management center, the server device 500 accumulates and stores the three-dimensional data and the prosthesis data acquired from each of the locals A to C, and holds them as big data.

Note that the server device 500 is not limited to being arranged in a management center such as a dental clinic, which is different from the local one, and may be arranged locally. For example, the server apparatus 500 may be arranged within any one of the locals A to C. FIG. Also, a plurality of data processing devices 100 may be arranged within one local, and furthermore, a server device 500 capable of communicating with the plurality of data processing devices 100 may be arranged within the one local. Moreover, the server device 500 may be implemented in the form of a cloud service.

In the dental laboratory, intraoral three-dimensional data and prosthesis data are aggregated from various locations such as local A to C. For this reason, intraoral three-dimensional data and prosthesis data held at the dental laboratory may be transmitted to the management center via the network 5, or may be transmitted to CD (Compact Disc) and USB (Universal Serial) data. Bus) memory or other removable disk 550 to the management center.

The intraoral three-dimensional data and the prosthesis data may also be sent from each of the locals A to C to the management center via the removable disk 550 without going through the network 5 . Also, intraoral three-dimensional data and prosthesis data may be sent to each other via the network 5 or the removable disk 550 between each of the locals A to C.

The generative model 1144a held by each of the local A to C data processors 100 may be shared among the local A to C data processors 100. FIG.

Also, the server device 500 may have the functions of the data processing device 100 . For example, each of the locals A to C transmits the acquired intraoral three-dimensional data to the server device 500, and the server device 500 receives each of the three-dimensional data received from each of the locals A to C and the generated Prosthesis data in each may be generated based on the model. The server device 500 may then transmit the prosthesis data to each local AC or dental lab. In this way, the data processing devices 100 of each of the local A to C may share the generation model held by the server device 500 in the form of cloud service. In this way, each of the locals A to C can cause the server device 500 to generate the prosthesis data simply by transmitting the three-dimensional data to the server device 500 .

[Hardware Configuration of Data Processor]
FIG. 8 is a schematic diagram showing the hardware configuration of the data processing device 100 according to this embodiment. Data processing apparatus 100 may be implemented by a general-purpose computer, or may be implemented by a computer dedicated to scanner system 10, for example.

As shown in FIG. 8, the data processing apparatus 100 includes, as main hardware elements, a scanner interface 102, a display interface 103, a peripheral device interface 105, a network controller 106, a media reader 107, and a PC display 108. , a memory 109 , a storage 110 , and an arithmetic device 130 .

The scanner interface 102 is an interface for connecting the three-dimensional scanner 200 and realizes input/output of data between the data processing device 100 and the three-dimensional scanner 200 .

The display interface 103 is an interface for connecting the display 300 and realizes input/output of data between the data processing device 100 and the display 300 . Display 300 is configured by, for example, an LCD (Liquid Crystal Display) or an organic ELD (Electroluminescence Display).

Peripheral device interface 105 is an interface for connecting peripheral devices such as keyboard 601 and mouse 602, and implements data input/output between data processing apparatus 100 and the peripheral devices.

The network controller 106 controls, via the network 5, the equipment located at the dental laboratory, the server equipment 500 located at the management center, the automated manufacturing equipment 600, and other data processing equipment 100 located locally. Send and receive data to and from each. The network controller 106 supports arbitrary communication methods such as Ethernet (registered trademark), wireless LAN (Local Area Network), and Bluetooth (registered trademark).

The media reader 107 writes and reads various data such as three-dimensional data and prosthesis data to and from the removable disk 550 .

A PC display 108 is a dedicated display for the data processing apparatus 100 . PC display 108 is configured by, for example, an LCD or an organic EL display. Although PC display 108 is separate from display 300 in the present embodiment, it may be shared with display 300 .

The memory 109 provides a storage area for temporarily storing program codes, work memory, etc. when the arithmetic unit 130 executes an arbitrary program. The memory 109 is composed of, for example, a volatile memory device such as a DRAM (Dynamic Random Access Memory) or SRAM (Static Random Access Memory).

The storage 110 provides storage areas for storing various data necessary for identification processing, learning processing, data generation processing, and the like. The storage 110 is composed of a non-volatile memory device such as a hard disk or SSD (Solid State Drive), for example.

Storage 110 includes scan information 112, extraction model 1142a, generation model 1144a, latent variable data 1143a, identification program 120, learning program 121, data processing program (data generation program) 125, An OS (Operating System) 127 is stored.

The scan information 112 includes intraoral three-dimensional data 122 acquired by the three-dimensional scanner 200, prosthesis data 123 generated by data generation processing based on the three-dimensional data 122, and identification processing for the prosthesis data 123. and the identification result 124 by. The prosthesis data 123 and the identification result 124 are stored in the storage 110 in association with the three-dimensional data 122 .

The identification program 120 is a program for executing identification processing. The learning program 121 is a program for executing learning processing of the generative model 1144a. The data processing program 125 is a program for executing data generation processing, and is also called a data generation program.

The latent variable data 1143a is data for specifying the latent variables extracted in the learning stage, and one or more latent variable data are accumulated in the storage 110. FIG. Note that latent variable data for specifying a latent variable may be accumulated in the storage 110 each time a latent variable is extracted not only in the learning stage but also in the practical stage. In this way, not only in the learning stage but also in the practical stage, it is possible to increase the latent variable data referred to when generating prosthesis data, and to improve the accuracy of prosthesis data generation.

The arithmetic device 130 is an arithmetic entity that executes various types of processing such as identification processing, learning processing, and data generation processing by executing various programs, and is an example of a computer. Arithmetic device 130 includes, for example, a CPU (Central Processing Unit) 132, an FPGA (Field-Programmable Gate Array) 134, a GPU (Graphics Processing Unit) 136, and the like.

[Accumulation of latent variable data]
An example of accumulation of latent variable data in the learning stage will be described with reference to FIGS. 9 and 10. FIG. 9 and 10 are schematic diagrams for explaining accumulation of latent variable data.

First, accumulation of latent variable data performed for each tooth will be described with reference to FIG. As shown in FIG. 9, in the learning stage, the extraction unit 1142 extracts latent variables relating to the characteristics of a given tooth from the three-dimensional data of that tooth. The latent variable data for specifying the latent variables extracted by the extraction unit 1142 are accumulated by the accumulation unit 1143 .

For example, from the three-dimensional data of the third molar (number 8) on the right side of the mandible, latent variables relating to the features of the third molar (number 8) are extracted and stored in the storage unit 1143 . From the second molar (number 7) on the right side of the mandible, latent variables relating to the characteristics of the second molar (number 7) are extracted and stored in the storage unit 1143 . From the first molar (sixth) on the right side of the mandible, latent variables relating to the features of the first molar (sixth) are extracted and accumulated in the accumulation unit 1143 . From the second premolar (number 5) on the right side of the mandible, latent variables relating to the characteristics of the second premolar (number 5) are extracted and stored in the storage unit 1143 . From the first premolar (number 4) on the right side of the mandible, latent variables relating to the features of the first premolar (number 4) are extracted and stored in the storage unit 1143 .

Similarly, latent variables are extracted for each tooth of the other teeth on the right side of the mandible and each tooth on the left side of the mandible and are accumulated in the accumulation unit 1143 . In addition, latent variables are extracted for each tooth and stored in the storage unit 1143 not only for the dental arch of the lower jaw but also for the dental arch of the upper jaw.

In this way, for each dental arch of the upper and lower jaws, the latent variables relating to the characteristics of each tooth are accumulated, with each tooth as an arbitrary tooth. For example, for each tooth, for each element that characterizes the tooth, such as the shape of the tooth (overall size, overall shape, edge shape, etc.), the position of the tooth, and the color of the tooth, there is a potential It is extracted as a variable and accumulated in the accumulation unit 1143 as latent variable data.

Next, accumulation of latent variable data performed for each of a plurality of teeth will be described with reference to FIG. As shown in FIG. 10, in the learning stage, the extraction unit 1142 extracts latent variables relating to the features of multiple teeth from the three-dimensional data of multiple teeth. The latent variable data for specifying the latent variables extracted by the extraction unit 1142 are accumulated by the accumulation unit 1143 .

For example, from the three-dimensional data of the first molar (6th), the second molar (7th), and the third molar (8th) on the right side of the mandible, the first molar (6th), the second A latent variable relating to the features of the molar (7th) and the third molar (8th) is extracted and stored in the storage unit 1143 . From the three-dimensional data of the second premolar (5th), first molar (6th), and second molar (7th) on the right side of the mandible, the second premolar (5th) and first molar (No. 6) and the second molar (No. 7) are extracted and accumulated in the accumulation unit 1143 . From the three-dimensional data of the first premolar (4th), the second premolar (5th), and the first molar (6th) on the right side of the mandible, the first premolar (4th) and the second premolar (No. 5) and the features of the first molar (No. 6) are extracted and accumulated in the accumulation unit 1143 . From the three-dimensional data of the mandibular right canine (3rd), first premolar (4th), and second premolar (5th), the corresponding canine (3rd), first premolar (4th), and A latent variable relating to the features of the second premolar (number 5) is extracted and stored in the storage unit 1143 . From the three-dimensional data of the lateral incisor (2nd), canine (3rd), and first premolar (4th) on the right side of the mandible, the lateral incisor (2nd), canine (3rd), and first premolar (4th) A latent variable relating to the feature of the premolar (number 4) is extracted and stored in the storage unit 1143 . From the three-dimensional data of the central incisor (1st), lateral incisor (2nd), and canine (3rd) on the right side of the mandible, the central incisor (1st), lateral incisor (2nd), and canine A latent variable related to the feature (3) is extracted and stored in the storage unit 1143 .

Similarly, latent variables are extracted for each of the plurality of teeth for the other teeth on the right side of the mandible and each tooth on the left side of the mandible and are accumulated in the accumulation unit 1143 . In addition, latent variables are extracted for each of a plurality of teeth and stored in the storage unit 1143 not only for the dental arch of the lower jaw but also for the dental arch of the upper jaw.

In this way, for each dental arch of the maxilla and mandible, a plurality of teeth are grouped together as an arbitrary tooth, and latent variables relating to the characteristics of the plurality of teeth are accumulated. For example, for each tooth, the shape of each tooth (overall size, overall shape, edge shape, etc.), the position of each tooth, the color of each tooth, and the relationship with surrounding teeth (positional relationship, degree of contact) etc.) are extracted as latent variables and stored in the storage unit 1143 as latent variable data.

In addition, it is not limited to an arbitrary tooth such as one tooth or a plurality of adjacent teeth, and the opposite tooth opposite to the arbitrary tooth, the tooth adjacent to the opposite tooth, and the opposite side of the arbitrary tooth ), latent variables may be similarly extracted and stored in the storage unit 1143 . In this way, the data processing apparatus 100 can generate prosthesis data suitable for any given tooth using latent variables relating to the features of teeth located around the given tooth.

[Example of latent variable data]
FIG. 11 is a schematic diagram showing an example of accumulated latent variable data. For the convenience of explanation, FIG. The more dimensions, the more dimensions the distribution of the latent variable data can be represented. In the figure, the latent variable represented by the horizontal axis is "Z(1)" and the latent variable represented by the vertical axis is "Z(2)".

As shown in FIG. 11, when two characteristic latent variables are selected from a plurality of latent variables and the latent variable data is mapped two-dimensionally, the arrangement of the tooth type (tooth number) is determined. Places tend to cluster. For example, latent variable data extracted from tooth No. 1 are represented by white circles, and each data is located at a short distance. The latent variable data extracted from tooth No. 2 are represented by rhombuses, and each data is located at a short distance. The latent variable data extracted from tooth No. 3 are represented by crosses, and each data is located at a short distance. The latent variable data extracted from tooth No. 4 are represented by black circles, and each data is located at a short distance. The latent variable data extracted from tooth No. 5 are represented by stars, and each data is located at a short distance. The latent variable data extracted from tooth No. 6 are represented by triangles, and each data is located at a short distance. The latent variable data extracted from tooth No. 7 are represented by white squares, and each data is located at a short distance. The latent variable data extracted from tooth No. 8 are represented by black squares, and each data is located at a short distance.

As described above, the shape is characteristic according to the type of tooth, and teeth of the same type have similar features. Therefore, the latent variable data extracted for each tooth also tend to have similar values among teeth of the same type.

[Generation of prosthesis data in the learning stage]
FIG. 12 is a schematic diagram showing an example of prosthesis data generation in the learning stage. In the example shown in FIG. 12, in the learning stage, the input unit 1102 inputs the three-dimensional data of each tooth in the tooth row in which the three-dimensional data of tooth No. 5 is omitted.

In the learning stage, the extraction unit 1142 extracts latent variables related to tooth features from the three-dimensional data acquired by the acquisition unit 1141 based on the extraction model 1142a. Specifically, the extraction unit 1142 extracts latent variables related to the characteristics of each of the teeth No. 4 and No. 6 adjacent to the tooth No. 5.

Here, as described with reference to FIGS. 9 to 11, the accumulation unit 1143 already accumulates a plurality of latent variable data. is mapped as a data distribution of Based on the generative model 1144a, the generation unit 1144 uses the latent variable data corresponding to the tooth No. 4 and the latent variable data corresponding to the tooth No. 6 to generate the latent variable data corresponding to the tooth No. 5. Based on the identified latent variable data corresponding to the fifth tooth, prosthesis data corresponding to the omitted fifth tooth is generated.

The prosthesis data generated in this manner undergoes identification processing by the identification unit 1106, and the generative model 1144a is machine-learned by feedback of the identification result.

[Generation of prosthesis data in the practical stage]
FIG. 13 is a schematic diagram showing an example of prosthesis data generation in the practical stage. In the example shown in FIG. 13, three-dimensional data of each tooth in a row of teeth in which tooth No. 5 is missing is input from the input unit 1102 in the practical stage.

In the practical stage, the extraction unit 1142 extracts latent variables related to tooth characteristics from the three-dimensional data acquired by the acquisition unit 1141 based on the extraction model 1142a. Specifically, the extraction unit 1142 extracts latent variables related to the characteristics of each of the teeth No. 4 and No. 6 adjacent to the missing tooth No. 5.

Based on the generative model 1144a, the generation unit 1144 uses the latent variable data corresponding to the tooth No. 4 and the latent variable data corresponding to the tooth No. 6 to generate the latent variable data corresponding to the tooth No. 5. Based on the latent variable data corresponding to the identified tooth No. 5, prosthetic data corresponding to the missing tooth No. 5 is generated.

In the above example, the generation unit 1144 generates latent variable data corresponding to tooth number 4 on the right side of the lower jaw as latent variable data corresponding to the "peripheral tooth" of missing tooth number 5 on the right side of the lower jaw, which is to be restored. , and latent variable data corresponding to tooth number 6 on the right side of the mandible, prosthesis data corresponding to missing tooth number 5 on the right side of the mandible are generated, but latent variable data corresponding to other peripheral teeth may be used to generate prosthesis data corresponding to missing tooth No. 5 on the right side of the mandible. For example, the generating unit 1144 identifies latent variable data corresponding to tooth number 5 on the right side of the upper jaw, which is the opposing tooth facing missing tooth number 5 on the right side of the lower jaw, and identifies the latent variable data corresponding to tooth number 5 on the right side of the upper jaw. Based on the latent variable data, prosthesis data corresponding to the missing tooth No. 5 on the right side of the mandible may be generated.

In this way, prosthetic data corresponding to the missing tooth is generated based on the three-dimensional data of the tooth row including the missing tooth and peripheral teeth of the missing tooth acquired by the three-dimensional scanner 200 .

[Flowchart of learning process]
FIG. 14 is a flowchart for explaining the learning process executed by data processing apparatus 100 according to the present embodiment. Each step shown in FIG. 14 is realized by the arithmetic device 130 of the data processing device 100 executing the OS 127, the identification program 120, the learning program 121, the data processing program (data generation program) 125, and the like. be.

As shown in FIG. 14, the data processing device 100 determines whether or not the conditions for starting the learning process are satisfied (S1). The start condition may be established, for example, when data processing device 100 performs some action for executing the learning process. For example, the starting condition may be that the learning start icon of the learning application has been operated by the user.

If the start condition is not met (NO in S1), the data processing device 100 ends this process. On the other hand, when the start condition is satisfied (YES in S1), the data processing device 100 determines whether or not three-dimensional data has been acquired (S2). For example, the data processing device 100 determines whether or not a sufficient amount of three-dimensional data has been acquired to generate prosthesis data. If the data processing device 100 has not acquired a sufficient amount of three-dimensional data (NO in S2), it repeats the processing of S2.

On the other hand, when a sufficient amount of three-dimensional data has been obtained (YES in S2), the data processing apparatus 100 extracts latent variables of peripheral teeth located around an arbitrary tooth from the three-dimensional data based on the extraction model 1142a. (S3). The three-dimensional data includes at least three-dimensional data of peripheral teeth located around an arbitrary tooth for which prosthesis data is to be generated.

Based on the generative model 1144a, the data processing apparatus 100 uses the extracted latent variables and the already accumulated latent variables to generate three-dimensional data (prosthesis data) corresponding to an arbitrary tooth (S4).

Next, the data processing device 100 compares the generated prosthesis data corresponding to an arbitrary tooth with the three-dimensional data corresponding to the arbitrary tooth, which is correct data (S5). If both match (YES in S6), the data processing device 100 terminates this process. The data processing apparatus 100 is not limited to the case where the generated prosthesis data and the correct data completely match, but when the degree of matching between the generated prosthesis data and the correct data exceeds a predetermined reference value, the data processing apparatus 100 You can judge that they match.

On the other hand, if the two do not match (NO in S6), the data processing device 100 adjusts the generative model 1144a (S7). For example, data processor 100 adjusts the neural network or parameters included in generative model 1144a to optimize generative model 1144a. After that, the data processing device 100 returns to S3.

[Flowchart of data generation processing]
FIG. 15 is a flowchart for explaining data generation processing executed by the data processing device according to the present embodiment. Each step shown in FIG. 15 is implemented by the arithmetic device 130 of the data processing device 100 executing the OS 127, the identification program 120, the data processing program (data generation program) 125, and the like.

As shown in FIG. 15, the data processing device 100 determines whether or not a condition for starting data generation processing is satisfied (S11). The start condition may be established, for example, when some action for executing data generation processing is performed in data processing device 100 . Specifically, the start condition may be satisfied when the power of the three-dimensional scanner 200 is turned on, or the mode corresponding to the data generation process may be established after the power of the three-dimensional scanner 200 is turned on. may be established at times. Alternatively, the start condition may be satisfied when the start switch is operated after an icon corresponding to the data generation process (for example, an AI assist icon) is operated and the icon flashes.

If the start condition is not satisfied (NO in S11), the data processing device 100 ends this process. On the other hand, when the start condition is satisfied (YES in S11), the data processing device 100 determines whether or not three-dimensional data has been obtained (S12). For example, the data processing device 100 determines whether or not a sufficient amount of three-dimensional data has been acquired to generate prosthesis data. If the data processing device 100 has not acquired a sufficient amount of three-dimensional data (NO in S12), it repeats the processing of S12.

On the other hand, when a sufficient amount of three-dimensional data is acquired (YES in S12), the data processing apparatus 100 extracts latent variables of peripheral teeth located around the missing tooth from the three-dimensional data based on the extraction model 1142a. (S13). The three-dimensional data includes at least three-dimensional data of peripheral teeth positioned around the missing tooth for which prosthesis data is to be generated.

Based on the generation model 1144a, the data processing apparatus 100 uses the extracted latent variables and the already accumulated latent variables to generate prosthesis data for prosthesis of the missing tooth (S14).

Next, the data processing device 100 outputs the generated prosthesis data to the dental laboratory, the automatic manufacturing device 600, or the server device 500 (S15). After that, the data processing device 100 terminates this process.

[Main configuration]
As described above, the present embodiment includes the following disclosures.

The data processing apparatus 100 includes an acquisition unit 1141 that acquires tooth three-dimensional data, and an extraction model 1142a that includes an algorithm subjected to machine learning to extract latent variables from a plurality of data. It comprises an extraction unit 1142 that extracts latent variables relating to tooth characteristics from the acquired three-dimensional data, and an accumulation unit 1143 that accumulates the latent variables extracted by the extraction unit 1142 .

As a result, the data processing apparatus 100 can extract and accumulate latent variables relating to tooth characteristics from the acquired three-dimensional data. Data (prosthesis data) can be generated. Therefore, by using the data processing device 100, the user 1 does not need to digitally design prosthesis data by himself/herself, and can more easily obtain an appropriate prosthesis.

The acquiring unit 1141 acquires at least one of an arbitrary tooth, a tooth adjacent to the arbitrary tooth, a tooth facing the arbitrary tooth, a tooth adjacent to the facing tooth, and a tooth opposite to the arbitrary tooth. Acquire one piece of three-dimensional data.

As a result, the data processing device 100 can extract and accumulate latent variables related to tooth characteristics from at least one of the three-dimensional data of adjacent teeth, opposing teeth, and teeth on the opposite side of an arbitrary tooth. Therefore, it is possible to generate prosthesis data suitable for an arbitrary tooth by taking into consideration the characteristics of peripheral teeth located around the arbitrary tooth.

The data processing apparatus 100 includes a generation unit 1144 that generates three-dimensional data of arbitrary teeth. The extraction unit 1142 extracts latent variables related to the features of the peripheral teeth from the three-dimensional data including at least the peripheral teeth, which are teeth positioned around an arbitrary tooth. A generation unit 1144 generates latent variables related to peripheral tooth features extracted by the extraction unit 1142 based on a generative model 1144a including an algorithm subjected to machine learning to generate data based on a plurality of latent variables, and accumulates Using the latent variables accumulated by the unit 1143, the three-dimensional data of the tooth of the same type as the arbitrary tooth type is generated.

As a result, the data processing device 100 uses the latent variables extracted from the three-dimensional data of the peripheral teeth located around the given tooth and the already accumulated latent variables to obtain a prosthesis that fits the given tooth. Since the data can be generated, it is possible to generate the prosthesis data that fits the arbitrary tooth considering the characteristics of the peripheral teeth located around the arbitrary tooth.

The data processing apparatus 100 determines whether or not the three-dimensional data of the same type of tooth generated by the generation unit 1144 is appropriate based on the three-dimensional data of the desired tooth input as the correct answer data. An identification unit 1106 for identification is provided. The generative model 1144a is machine-learned based on the identification result of the identification unit 1106 .

As a result, the data processing apparatus 100 can machine-learn the generative model 1144a by identifying whether or not the generated three-dimensional data (prosthesis data) is appropriate, thereby improving the prosthesis data generation accuracy. can be made

The data generation device (data processing device 100) includes an acquisition unit 1141 that acquires three-dimensional data including at least peripheral teeth that are teeth located around the missing tooth, and extracts latent variables from a plurality of data. An extraction unit 1142 for extracting latent variables related to peripheral tooth features from the three-dimensional data acquired by the acquisition unit 1141 based on an extraction model 1142a including an algorithm subjected to machine learning to Peripheral teeth extracted by the extraction unit 1142 based on a storage unit 1143 that stores a plurality of latent variables related to and the latent variables accumulated by the accumulation unit 1143 to generate prosthesis data for producing a prosthesis suitable for the missing part of the missing tooth.

As a result, the data generation device (data processing device 100) extracts latent variables related to the features of the peripheral teeth from the acquired three-dimensional data, and uses the extracted latent variables and the already accumulated latent variables to It is possible to generate prosthesis data for producing a prosthesis that fits the missing part of the tooth. Therefore, the user 1 does not need to digitally design prosthesis data by himself and can more easily obtain an appropriate prosthesis.

The scanner system 10 includes a three-dimensional scanner 200 that acquires three-dimensional data including at least peripheral teeth that are teeth located around the missing tooth using a three-dimensional camera, and the three-dimensional scanner 200 that acquires three-dimensional data. a data generation device (data processing device 100) for generating prosthesis data for manufacturing a prosthesis that fits the missing part of the missing tooth based on the obtained three-dimensional data. The data generation device (data processing device 100) includes an acquisition unit 1141 that acquires three-dimensional data including at least peripheral teeth, and an extraction model 1142a that includes an algorithm subjected to machine learning to extract latent variables from a plurality of data. Based on the three-dimensional data acquired by the acquisition unit 1141, the extraction unit 1142 extracts latent variables related to the features of the peripheral teeth, the storage unit 1143 stores a plurality of latent variables related to the features of the teeth in advance, and a plurality of Based on a generative model 1144a including an algorithm subjected to machine learning to generate data based on the latent variables, the latent variables related to the characteristics of the peripheral teeth extracted by the extraction unit 1142 and accumulated by the accumulation unit 1143. and a generation unit 1144 that generates prosthesis data for manufacturing a prosthesis that fits the missing part of the missing tooth, using the latent variables.

As a result, the data generation device (data processing device 100) of the scanner system 10 extracts latent variables related to the features of the peripheral teeth from the acquired three-dimensional data, and compares the extracted latent variables with the already accumulated latent variables. can be used to generate prosthesis data for producing a prosthesis that fits the missing part of the missing tooth. Therefore, by using the data generation device (data processing device 100), the user 1 does not need to digitally design prosthesis data by himself/herself, and can more easily obtain an appropriate prosthesis.

The data processing method comprises a step (S2) of acquiring three-dimensional data including at least peripheral teeth, which are teeth located around an arbitrary tooth, and machine learning was performed to extract latent variables from a plurality of data. A step of extracting latent variables related to peripheral tooth features from the acquired three-dimensional data based on an extraction model 1142a containing an algorithm (S3), and machine learning is performed to generate data based on the plurality of latent variables. A step of generating three-dimensional data of an arbitrary tooth based on the generative model 1144a including the extracted algorithm, using latent variables related to the extracted peripheral tooth features and accumulated latent variables related to tooth features. (S4), based on the three-dimensional data of the same tooth type as the arbitrary tooth type input as the correct answer data, whether or not the generated three-dimensional data of the same tooth type is appropriate. and a step (S5) of identifying whether the Generative model 1144a is machine-learned based on the identification result of the identifying step.

As a result, the latent variables related to the features of the surrounding teeth are extracted from the acquired three-dimensional data, and the extracted latent variables and the already accumulated latent variables are used to generate prosthesis data suitable for any tooth. be able to. Therefore, the user 1 does not need to digitally design prosthesis data by himself and can more easily obtain an appropriate prosthesis. Furthermore, by machine-learning the generation model 1144a using the identification result of whether or not the prosthesis data generated based on the acquired three-dimensional data is appropriate, more appropriate prosthesis data can be generated. Become.

The data processing program causes the computer (arithmetic device 130) to acquire three-dimensional data including at least peripheral teeth, which are teeth located around an arbitrary tooth (S2), and extract latent variables from a plurality of data. A step (S3) of extracting latent variables related to the features of peripheral teeth from the acquired three-dimensional data based on an extraction model 1142a including an algorithm subjected to machine learning to extract data based on a plurality of latent variables Based on a generative model 1144a that includes an algorithm subjected to machine learning to generate an arbitrary a step (S4) of generating three-dimensional data of a tooth of the same type as the type of tooth of the generated tooth of the same type as the arbitrary tooth type based on the three-dimensional data of the arbitrary tooth input as the correct answer data; and a step (S5) of discriminating whether or not the three-dimensional data of the teeth are appropriate. Generative model 1144a is machine-learned based on the identification result of the identifying step.

As a result, the latent variables related to the features of the surrounding teeth are extracted from the acquired three-dimensional data, and the extracted latent variables and the already accumulated latent variables are used to generate prosthesis data suitable for any tooth. be able to. Therefore, the user 1 does not need to digitally design prosthesis data by himself and can more easily obtain an appropriate prosthesis. Furthermore, by machine-learning the generation model 1144a using the identification result of whether or not the prosthesis data generated based on the acquired three-dimensional data is appropriate, more appropriate prosthesis data can be generated. Become.

The data generation method includes a step of acquiring three-dimensional data including at least peripheral teeth, which are teeth located around the missing tooth (S12), and machine learning for extracting latent variables from a plurality of data. (S13) for extracting latent variables relating to features of peripheral teeth from the acquired three-dimensional data based on an extraction model 1142a including an algorithm performed by; Based on the generative model 1144a including the machine-learned algorithm, the missing part of the missing tooth is fitted using the latent variables relating to the extracted peripheral tooth features and the accumulated latent variables relating to the tooth features. generating (S14) prosthesis data for fabricating a prosthesis that

As a result, the latent variables related to the characteristics of the surrounding teeth are extracted from the acquired three-dimensional data, and the extracted latent variables and the already accumulated latent variables are used to generate prosthesis data that fits the missing part of the missing tooth. can be generated. Therefore, the user 1 does not need to digitally design prosthesis data by himself and can more easily obtain an appropriate prosthesis.

The data generation program causes the computer (arithmetic device 130) to acquire three-dimensional data including at least peripheral teeth, which are teeth located around the missing tooth (S12), and obtain three-dimensional data from a plurality of data a step of extracting latent variables related to the characteristics of peripheral teeth from the acquired three-dimensional data based on an extraction model 1142a including an algorithm subjected to machine learning for extracting latent variables (S13); Based on a generative model 1144a that includes a machine-learned algorithm to generate data based on variables, latent variables for extracted peripheral tooth features and latent variables for accumulated tooth features are used. Then, a step (S14) of generating prosthesis data for producing a prosthesis suitable for the missing part of the missing tooth is executed.

As a result, the latent variables related to the characteristics of the surrounding teeth are extracted from the acquired three-dimensional data, and the extracted latent variables and the already accumulated latent variables are used to generate prosthesis data that fits the missing part of the missing tooth. can be generated. Therefore, the user 1 does not need to digitally design prosthesis data by himself and can more easily obtain an appropriate prosthesis.

[Modification]
The present invention is not limited to the above embodiments, and various modifications and applications are possible. Modifications applicable to the present invention will be described below.

In the data processing device 100 (data generation device), attribute data of the subject 2 who is the owner of the dentition may be input together with the three-dimensional data of the dentition. The attribute data includes attribute information (profile) regarding the subject 2, such as age, sex, race, height, weight, and place of residence. Furthermore, the data processing device 100 may generate prosthesis data considering attribute information (profile) about the subject 2 .

In general, tooth shape differs depending on genetics or living environment such as age, sex, race, height, weight, place of residence, and the like. For example, the permanent teeth of adults are generally larger than the milk teeth of children, and the two have different shapes. In general, male teeth are larger than female teeth, and the two have different shapes. In general, the teeth of Westerners tend to have sharp tips so that they can easily bite off hard meat and bread, whereas the teeth of Japanese people tend to have sharp tips so that they can easily grind soft rice and vegetables. It tends to be smooth. For this reason, if the generative model 1144a is machine-learned based on the attribute information (profile), a prosthesis for producing a prosthesis suitable for the attribute information (profile) regarding the subject 2 in consideration of heredity, living environment, etc. data can be generated.

Further, in the data processing device 100 (data generation device), along with the three-dimensional data of the row of teeth, motion data relating to occlusal motion of at least one of adjacent teeth and opposing teeth may be input. The bite movement includes, for example, vertical movement of the mandible for crushing food, back and forth movement of the mandible for crushing food, and the like. Furthermore, the data processing device 100 may extract latent variables for each tooth in consideration of occlusal motion in at least one of adjacent teeth and opposing teeth, and use the extracted latent variables to generate prosthetic data. may be generated.

In this way, the input unit 1102 may receive motion data of the jaw when the jaw moves. In this way, in the learning stage, latent variables are accumulated in consideration of jaw movement data. Furthermore, in the practical stage, jaw motion data may be input in addition to the three-dimensional data of peripheral teeth located around the missing tooth. In this way, since the shape of the prosthesis can be determined in consideration of the movement data of the jaw, the user 1 can obtain a more suitable prosthesis.

It should be considered that the embodiments disclosed this time are illustrative in all respects and not restrictive. The scope of the present invention is indicated by the scope of the claims rather than the above description, and is intended to include all modifications within the scope and meaning equivalent to the scope of the claims. Note that the configurations exemplified in this embodiment and the configurations exemplified in the modifications can be combined as appropriate.

1 user, 2 subject, 5 network, 10 scanner system, 100 data processor, 102 scanner interface, 103 display interface, 105 peripheral device interface, 106 network controller, 107 media reader, 108, 300 display, 109 memory, 110 storage, 112 scan information, 120 identification program, 121 learning program, 122 three-dimensional data, 123 prosthesis data, 124 identification result, 125 data processing program, 130 computing device, 200 three-dimensional scanner, 500 server device, 550 Removable disk, 600 automatic manufacturing device, 601 keyboard, 602 mouse, 1102 input unit, 1104 data generation unit, 1106 identification unit, 1141 acquisition unit, 1142 extraction unit, 1142a extraction model, 1143 storage unit, 1143a latent variable data, 1144 generation Part, 1144a Generative model.

Claims (7)

A data generation device for generating three-dimensional data for producing a dental prosthesis ,
For each of a plurality of teeth in which the missing tooth, which is a missing tooth, and the surrounding teeth that are located around the missing tooth and do not include the missing tooth, are adjacent to each other, the tertiary of the scan target region scanned in the adjacent teeth an acquisition unit for acquiring three-dimensional data including three-dimensional position information at each of a plurality of points constituting the scan target region for indicating the original shape ;
Latent variables related to features of the peripheral teeth from the three- dimensional data acquired by the acquisition unit based on an extraction model including a first algorithm in VAE in which machine learning is performed to extract latent variables from a plurality of data an extracting unit that extracts for each of the plurality of adjacent teeth;
an accumulation unit for accumulating in advance latent variables relating to tooth characteristics for each of the plurality of adjacent teeth ;
of the peripheral teeth for each of the adjacent teeth extracted by the extraction unit based on a generative model including a second algorithm in VAE in which machine learning is performed to generate data based on a plurality of latent variables; For producing the prosthesis that fits the missing part of the missing tooth using the latent variables relating to the features and the latent variables relating to the features of the adjacent teeth for each of the adjacent teeth accumulated by the storage unit. A data generation device comprising a generation unit that generates three-dimensional data . an identification unit that identifies whether or not the three-dimensional data for fabricating the prosthesis generated by the generation unit is appropriate based on the three-dimensional data of the missing tooth input as correct data;
The data generating device according to claim 1 , wherein the generative model is machine-learned based on a result of identification by the identification unit. A scanner system for generating three-dimensional data for producing a dental prosthesis ,
Using a three-dimensional camera, for each of the adjacent teeth between the missing tooth and the surrounding teeth that are located around the missing tooth and do not include the missing tooth , the adjacent teeth are scanned. a three-dimensional scanner for acquiring three-dimensional data including three-dimensional positional information at each of a plurality of points forming the scanned target region for indicating the three-dimensional shape of the scanned target region ;
a data generation device for generating three- dimensional data for manufacturing the prosthesis that fits the missing part of the missing tooth based on the three- dimensional data acquired by the three-dimensional scanner;
The data generation device is
an acquisition unit that acquires three-dimensional data including three-dimensional position information at each of a plurality of points that constitute the scan target region acquired by the three-dimensional scanner;
Latent variables related to features of the peripheral teeth from the three- dimensional data acquired by the acquisition unit based on an extraction model including a first algorithm in VAE in which machine learning is performed to extract latent variables from a plurality of data an extracting unit that extracts for each of the plurality of adjacent teeth;
an accumulation unit for accumulating in advance latent variables relating to tooth characteristics for each of the plurality of adjacent teeth;
of the peripheral teeth for each of the adjacent teeth extracted by the extraction unit based on a generative model including a second algorithm in VAE in which machine learning is performed to generate data based on a plurality of latent variables; a generating unit that generates three-dimensional data for manufacturing the prosthesis by using the latent variables related to the features and the latent variables related to the tooth features of the plurality of adjacent teeth accumulated by the storage unit; scanner system, including A data generation method in which a computer generates three-dimensional data for producing a dental prosthesis,
As a process executed by the computer,
For each of a plurality of teeth in which the missing tooth , which is a missing tooth, and the surrounding teeth that are located around the missing tooth and do not include the missing tooth , are adjacent to each other, the tertiary of the scan target region scanned in the adjacent teeth obtaining three-dimensional data including three-dimensional position information at each of a plurality of points constituting the scan target region for indicating the original shape ;
Based on an extraction model that includes a first algorithm in VAE that is machine-learned to extract latent variables from a plurality of data, from the three- dimensional data acquired by the acquiring step , latent variables for features of the peripheral teeth. extracting a variable for each of the adjacent teeth;
the peripheral teeth for each of the adjacent teeth extracted by the extracting step based on a generative model comprising a second algorithm in VAE subjected to machine learning to generate data based on a plurality of latent variables; and a latent variable related to the tooth characteristics of each of the adjacent teeth that have been accumulated in advance to create the prosthesis that fits the missing part of the missing tooth. A method of generating data, comprising the step of generating data. further comprising a step of identifying whether or not the three-dimensional data for fabricating the prosthesis generated by the generating step is appropriate, based on the three-dimensional data of the missing tooth input as correct data. Item 5. The data generation method according to item 4. A data generation program for generating three-dimensional data for manufacturing a dental prosthesis,
The data generation program is a computer,
For each of a plurality of teeth in which the missing tooth , which is a missing tooth, and the surrounding teeth that are located around the missing tooth and do not include the missing tooth , are adjacent to each other, the tertiary of the scan target region scanned in the adjacent teeth obtaining three-dimensional data including three-dimensional position information at each of a plurality of points constituting the scan target region for indicating the original shape ;
Based on an extraction model that includes a first algorithm in VAE that is machine-learned to extract latent variables from a plurality of data, from the three- dimensional data acquired by the acquiring step , latent variables for features of the peripheral teeth. extracting a variable for each of the adjacent teeth;
the peripheral teeth for each of the adjacent teeth extracted by the extracting step based on a generative model comprising a second algorithm in VAE subjected to machine learning to generate data based on a plurality of latent variables; and a latent variable related to the tooth characteristics of each of the adjacent teeth that have been accumulated in advance to create the prosthesis that fits the missing part of the missing tooth. A data generation program that causes the step of generating data to be performed. further comprising a step of identifying whether or not the three-dimensional data for fabricating the prosthesis generated by the generating step is appropriate, based on the three-dimensional data of the missing tooth input as correct data. 7. A data generation program according to item 6.

Images