Support device, support method, and support program

Abstract

To provide a technique capable of diagnosing a disease state in a biological tissue in the oral cavity in a low invasive manner in a short time.SOLUTION: A support device 1 includes: an input interface 14 to which image data representing the three-dimensional shape of a biological tissue is input; and a calculation device 11 which derives support information by using the image data input from the input interface 14 and an estimation model 50 that has been trained by machine learning so as to derive support information including at least information about the relative positions of the tooth and the gingiva for supporting diagnosis of a disease state in the biological tissue based on the image data.SELECTED DRAWING: Figure 2

Description

Technical Field

[0001] The present disclosure relates to a support device, a support method, and a support program for assisting in the diagnosis of pathological conditions in living tissues in the oral cavity, including at least teeth and gums.

Background Art

[0002] Periodontal disease is an infectious inflammatory disease caused by periodontal pathogenic bacteria. Periodontal disease includes diseases that occur in periodontal tissues composed of gums, cementum, periodontal ligament, and alveolar bone, necrotizing periodontal diseases, periodontal abscesses, periodontal-endodontic lesions, gingival recession (e.g., the gums recede), alveolar bone recession (e.g., the alveolar bone recedes), and occlusal trauma caused by strong occlusal force or abnormal force. When periodontal disease progresses, inflammation of the gums, tooth mobility, or alveolar bone recession occurs.

[0003] As a method for diagnosing pathological conditions in living tissues in the oral cavity such as teeth and gums, a method of measuring the depth of periodontal pockets is known. For example, Patent Document 1 discloses that an operator such as a dentist inserts a probe of a handpiece into a periodontal pocket to measure the depth of the periodontal pocket, thereby checking the state of teeth and gums.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] In the periodontal disease inspection method disclosed in Patent Document 1, since it is necessary to insert a probe into the periodontal pocket, the burden on the patient is large, and the inspection result of periodontal disease may vary depending on the skill of the operator who measures the depth of the periodontal pocket. Since the probe is inserted into and removed from the periodontal pocket where pathogenic bacteria exist, if pathogenic bacteria invade the tooth through the probe in a tooth with mild or no symptoms, there is a risk that the tooth will be infected with periodontal disease. In addition, pain or bleeding may occur in the patient when measuring the depth of the periodontal pocket, and when bleeding occurs, there is also a possibility that pathogenic bacteria will enter the blood. Furthermore, when measuring the depth of the periodontal pocket at a plurality of locations for each tooth, the measurement time may become long.

[0006] The present disclosure has been made to solve such problems, and an object thereof is to provide a technique capable of diagnosing the pathological condition in a living tissue in the oral cavity with low invasiveness and in a short time.

Means for Solving the Problems

[0007] According to an example of the present disclosure, a support device for supporting the diagnosis of a pathological condition in a living tissue in the oral cavity including at least teeth and gums is provided. The support device includes an input unit into which image data showing the three-dimensional shape of the living tissue Input data including the measurement points of the biological tissue designated for diagnosing the pathological condition and the measurement directions at the measurement points is input, Input data, Input and an arithmetic unit that derives support information using an estimation model trained by machine learning so as to support the diagnosis of the pathological condition in the living tissue based on the of the support data. The calculation unit derives position information corresponding to each of a plurality of levels indicating the positions of the biological tissue existing along the measurement direction at the measurement point, calculates the distances between the plurality of levels based on the position information corresponding to each of the plurality of levels, and calculates at least one of information indicating the type of pathological condition and information indicating the degree of progression of the pathological condition as support information based on the distances between the plurality of levels is.

[0008] According to an example of the present disclosure, a support method for supporting the diagnosis of a pathological condition in a living tissue in the oral cavity including at least teeth and gums by a computer is provided. The support method includes, as a process executed by the computer, a step of acquiring image data showing the three-dimensional shape of the living tissue Input data including the measurement points of the biological tissue designated for diagnosing the pathological condition and the measurement directions at the measurement points and, Input data, Input and based on the of the supportincluding a step of deriving support information using an estimation model trained by machine learning to derive support information. The deriving step includes a step of deriving position information corresponding to each of a plurality of levels indicating the positions of the biological tissue existing along the measurement direction at the measurement point, a step of calculating the distances between the plurality of levels based on the position information corresponding to each of the plurality of levels, and a step of calculating at least one of information indicating the type of pathological condition and information indicating the degree of progression of the pathological condition as support information based on the distances between the plurality of levels

[0009] According to an example of the present disclosure, a support program for assisting in the diagnosis of a pathological condition in a biological tissue in the oral cavity including at least teeth and gums by a computer is provided. The support program causes the computer to Input data including the measurement points of the biological tissue designated for diagnosing the pathological condition and the measurement directions at the measurement points acquire image data showing the three-dimensional shape of the biological tissue, Input data, and Input using an estimation model trained by machine learning to derive support information for assisting in the diagnosis of a pathological condition in the biological tissue based on the data, of the support derive support information. and The deriving step includes a step of deriving position information corresponding to each of a plurality of levels indicating the positions of the biological tissue existing along the measurement direction at the measurement point, a step of calculating the distances between the plurality of levels based on the position information corresponding to each of the plurality of levels, and a step of calculating at least one of information indicating the type of pathological condition and information indicating the degree of progression of the pathological condition as support information based on the distances between the plurality of levels

Advantages of the Invention

[0010] According to the present disclosure, by using an estimation model, support information for assisting in the diagnosis of a pathological condition in a biological tissue in the oral cavity including at least teeth and gums can be derived based on image data showing the three-dimensional shape of the biological tissue, so that the pathological condition in the biological tissue in the oral cavity can be diagnosed with low invasiveness and in a short time.

Brief Description of the Drawings

[0011]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

Figure 13

Figure 14

Figure 15

Figure 16

Figure 17

Figure 18

Figure 19

Figure 20

Figure 21

Figure 22

Figure 23

Figure 24

Embodiments for Carrying Out the Invention

[0012] <Embodiment 1> Embodiment 1 of the present disclosure will be described in detail with reference to the drawings. For the same or corresponding parts in the drawings, the same reference numerals are given and the description thereof will not be repeated.

[0013] [Application Example] With reference to FIG. 1, an application example of the support device 1 according to Embodiment 1 will be described. FIG. 1 is a diagram showing an application example of the support device 1 according to Embodiment 1.

[0014] As the most commonly used method in the examination of periodontal disease, there is a method of measuring the depth of the periodontal pocket. Specifically, an operator such as a dentist inserts a mechanical or electrical probe into the periodontal pocket existing at at least one measurement point set around the tooth, and measures the depth from the gingival margin to the gingival boundary (the bottom of the periodontal pocket), which is the part where the gingiva is normally attached to the tooth, to measure the depth of the periodontal pocket. The number of measurement points can be set arbitrarily, such as 4 points or 6 points, for one tooth. In the example of FIG. 1, 6 measurement points are set for one tooth.

[0015] In addition to the method of measuring the depth of the periodontal pocket, examples of the periodontal disease examination methods used other than this include methods using X-ray images obtained by photographing the oral cavity. This method measures the distance between the crown tip and the alveolar bone crest, and the distance between the alveolar bone crest and the root apex based on the X-ray image of the oral cavity, and calculates the ratio of these (hereinafter also referred to as the "crown-root ratio") to measure the degree of alveolar bone recession.

[0016] As described above, the operator can perform a periodontal disease examination by measuring the periodontal pocket and using an X-ray image. However, from the viewpoints of the burden on the patient, variations in examination accuracy depending on the skill of the operator, and the lengthening of the examination time, these examination methods may not be preferable as a periodontal disease examination.

[0017] For example, in the periodontal disease examination method by measuring the periodontal pocket, since it is necessary to insert a probe into the periodontal pocket, the burden on the patient is large. Also, the examination results of periodontal disease may vary depending on the skill of the operator measuring the depth of the periodontal pocket, and it requires rich experience and time for the operator to diagnose multiple periodontal disease diagnosis parameters based on experience and predict the progression of periodontal disease. Since the probe is inserted into and removed from the periodontal pocket where pathogenic bacteria exist, if pathogenic bacteria invade a tooth in a state with mild or no symptoms via the probe, there is a risk that the tooth will be infected with periodontal disease. In addition, pain or bleeding may occur in the patient when measuring the depth of the periodontal pocket, and when bleeding occurs, there is also a possibility that pathogenic bacteria will enter the blood. Furthermore, when measuring the depth of the periodontal pocket at multiple locations for each tooth, the measurement time may become long.

[0018] In the periodontal disease examination method using an X-ray image, intraoral X-ray photography or panoramic photography is performed, but only the state of the tooth as seen from the X-ray incident direction can be observed, and since an image superimposed in the direction in which the X-ray is incident is shown, it is difficult to accurately measure the crown-root ratio.

[0019] For the reasons described above, when formulating future treatment strategies, it is difficult to quantitatively accumulate data on the test results of periodontal disease, and it is also difficult to estimate the degree of progression of periodontal disease over time. Furthermore, when explaining the state of periodontal disease to patients, there is a need for technology that can obtain informed consent in a short period of time and gain the understanding of patients.

[0020] Therefore, the support device 1 according to Embodiment 1 is configured to estimate (derive) support information for assisting in the diagnosis of pathological conditions in a living tissue based on image data showing the three-dimensional shape of the living tissue in the oral cavity including at least teeth and gums, using AI (Artificial Intelligence) technology. Note that the living tissue is an object in the oral cavity including at least teeth and gums and does not include artificial objects such as implants.

[0021] Specifically, the user of the support device 1 acquires three-dimensional data (optical scanner data) including the position information of each point of a point cloud (a plurality of points) showing the surface of a living tissue including teeth and gums in the oral cavity by scanning the oral cavity of a patient using a three-dimensional scanner (optical scanner) (not shown). The three-dimensional data includes, as position information, the coordinates (X, Y, Z) of each point showing the surface of the living tissue in a predetermined horizontal direction (X-axis direction), vertical direction (Y-axis direction), and height direction (Z-axis direction). Furthermore, the three-dimensional data may include color information indicating the actual color of a portion (the surface portion of the living tissue) corresponding to each point of the point cloud (a plurality of points) showing the surface of the living tissue including teeth and gums in the oral cavity.

[0022] Note that the "user" includes operators (such as doctors) or assistants (such as dental assistants, dental technicians, nurses, etc.) in various fields such as dentistry, oral surgery, plastic surgery, reconstructive surgery, and cosmetic surgery. Also, the "patient" includes patients in dentistry, oral surgery, plastic surgery, reconstructive surgery, and cosmetic surgery. The three-dimensional scanner is a so-called intraoral scanner (IOS: Intra Oral Scanner) that can optically image the inside of a patient's mouth by methods such as the confocal method or the triangulation method, and can acquire the position information of each point of the point cloud that constitutes the surface of a biological tissue (for example, teeth and gums in the mouth) that is the object to be scanned placed in a certain coordinate space. By using the three-dimensional data acquired by the three-dimensional scanner, the user can generate a rendering image (appearance image) showing the three-dimensional shape of the biological tissue. The "rendering image" is an image generated by processing or editing certain data. For example, the user can generate a rendering image showing a two-dimensional biological tissue (a part of the biological tissue that can be shown by IOS data) viewed from a predetermined viewpoint by processing or editing the three-dimensional data of the biological tissue acquired by the three-dimensional scanner, and further, by changing the predetermined viewpoint in multiple directions, a plurality of rendering images showing the two-dimensional biological tissue (a part of the biological tissue that can be shown by IOS data) viewed from multiple directions can be generated.

[0023] In addition, the user obtains three-dimensional volume (voxel) data of hard tissue parts (bones, teeth, etc.) including soft tissue parts (skin, gums, etc.) around the upper and lower jaws of a patient by photographing the upper and lower jaws of the patient using a CT (Computed Tomography) imaging device (not shown). Note that the soft tissue parts are less detectable by X-rays and have lower data acquisition accuracy compared to the hard tissue parts. For this reason, when the soft tissue parts are displayed in an image, there is a possibility that the soft tissue parts can be faintly recognized or that parts that cannot be recognized may occur. The CT imaging device is an X-ray imaging device that performs computed tomography of the upper and lower jaws of a patient by rotating an X-ray transmitter and receiver, which are a type of radiation, around the patient's face. The user can generate a rendering image (tomographic image or external appearance image) showing the three-dimensional shape of the living tissue by using the volume data of the living tissue to be imaged obtained by the CT imaging device. For example, the user can generate a rendering image showing a two-dimensional living tissue (a part of the living tissue that can be shown by the CT data) viewed from a predetermined viewpoint by processing or editing the volume data of the living tissue obtained by the CT imaging device, and further, by changing the predetermined viewpoint in multiple directions, a plurality of rendering images showing the two-dimensional living tissue (a part of the living tissue that can be shown by the CT data) viewed from multiple directions can be generated.

[0024] Hereinafter, the three-dimensional data including the position information of each point of the point cloud indicating the surface of the biological tissue obtained by the three-dimensional scanner is also referred to as "IOS data", and the rendering image generated based on the IOS data is also referred to as "IOS image". Further, the three-dimensional volume data obtained by the CT imaging device is also referred to as "CT data", and the rendering image generated based on the CT data is also referred to as "CT image". The IOS image can show the surface shape of the biological tissue to be scanned in great detail, but it cannot show at all the internal structures (such as alveolar bone and root apex) that do not appear on the surface of the biological tissue. The CT image can show the hard tissue parts (bones, teeth, etc.) of the imaging target in relatively detail, but it cannot show the soft tissue parts (skin, gingiva, etc.) more detailed than the hard tissue parts.

[0025] The user can generate synthetic image data by synthesizing IOS data and CT data obtained for the same patient. Here, the IOS data and the CT data have different data formats from each other. Therefore, for example, the user can convert the data format of the IOS data to the data format of the CT data, and use the two converted data to perform pattern matching on the three-dimensional shape of the surface of the biological tissue, thereby generating synthetic image data obtained by synthesizing the IOS data and the CT data. Note that the user may also generate synthetic image data by converting the data format of the CT data to the data format of the IOS data and performing pattern matching on the three-dimensional shape of the surface of the biological tissue using the two converted data. Alternatively, the user may convert the data formats of the CT data and the IOS data to a common data format, and use the two converted data to perform pattern matching on the three-dimensional shape of the surface of the biological tissue, thereby generating synthetic image data. The user can generate a rendering image (for example, the synthetic image shown in FIG. 1) that shows a two-dimensional biological tissue (a part of the biological tissue that can be shown by both the IOS data and the CT data) viewed from a predetermined viewpoint by processing or editing the synthetic image data. As shown in FIG. 1, the synthetic image can three-dimensionally show the surface shape of the biological tissue shown by the IOS data and the tomographic structure or appearance in the hard tissue part (bones, teeth, etc.) shown by the CT data. Note that when generating the synthetic image data, the user may adjust the luminance, contrast, transparency, etc. as necessary in each of the IOS data and the CT data. Furthermore, the user may generate the synthetic image data after segmenting each of the plurality of teeth, the jawbone, and the alveolar bone in each of the IOS data and the CT data in advance.

[0026] Note that the support device 1 may acquire IOS data from a three-dimensional scanner, acquire CT data from a CT imaging device, and generate composite image data based on the acquired IOS data and CT data according to user input. Alternatively, the support device 1 may acquire composite image data generated by the user using another device from the other device without acquiring the IOS data and CT data.

[0027] In the composite image that can be generated based on the composite image data, the three-dimensional shape of hard tissue parts such as the alveolar bone and the root apex is shown by the CT data, and for the three-dimensional shape of soft tissue parts such as the gingiva that cannot be shown by the CT data, it can be shown by the IOS data. Thereby, the composite image can supplement soft tissue parts such as the gingiva that cannot be represented only by the CT data with the IOS data, and can show them in detail together with hard tissue parts such as the alveolar bone and the root apex.

[0028] Although it will be described in detail later, the support device 1 sets a predetermined measurement direction and a predetermined measurement point for measuring the depth of the periodontal pocket in the composite image data of biological tissue including at least teeth and gingiva. The support device 1 uses an estimation model 50 described later to derive support information including information regarding at least the relative positions of the teeth and the gingiva for supporting the diagnosis of the pathological condition in the biological tissue.

[0029] As shown in FIG. 1, for example, the support device 1 shows, as support information, the depth of the periodontal pocket at the set measurement point for each tooth shown by the composite image data. Thereby, since the support device 1 can derive support information for supporting the diagnosis of the pathological condition in the biological tissue based on the composite image data showing the three-dimensional shape of the biological tissue including at least teeth and gingiva, the user can diagnose the pathological condition in the oral biological tissue with low invasiveness and in a short time.

[0030] [Hardware Configuration of the Identification Device] While referring to FIG. 2, the hardware configuration of the support device 1 according to Embodiment 1 will be described. FIG. 2 is a block diagram showing the hardware configuration of the support device 1 according to Embodiment 1. The support device 1 may be realized by, for example, a general-purpose computer or a dedicated computer for a system for estimating support information.

[0031] As shown in FIG. 2, the support device 1 includes, as main hardware elements, an arithmetic unit 11, a memory 12, a storage device 13, an input interface 14, a display interface 15, a peripheral device interface 16, a media reader 17, and a communication device 18.

[0032] The arithmetic unit 11 is an arithmetic entity (computer) that executes various processes by executing various programs, and is an example of an "arithmetic unit". The arithmetic unit 11 is composed of a processor such as a CPU (central processing unit) or an MPU (Micro-processing unit), for example. Note that a processor, which is an example of the arithmetic unit 11, has a function of executing various processes by executing a program, but a part or all of these functions may be implemented using a dedicated hardware circuit such as an ASIC (Application Specific Integrated Circuit) or an FPGA (Field-Programmable Gate Array). The "processor" is not limited to a narrow sense processor that executes processing in a stored program manner like a CPU or an MPU, and may include a hardwired circuit such as an ASIC or an FPGA. Therefore, the "processor", which is an example of the arithmetic unit 11, can also be read as a processing circuitry whose processing is defined in advance by computer-readable code and / or a hardwired circuit. Note that the arithmetic unit 11 may be composed of one chip or a plurality of chips. Furthermore, the processor and related processing circuits may be composed of a plurality of computers interconnected by wire or wirelessly via a local area network or a wireless network, for example. The processor and related processing circuits may be composed of a cloud computer that remotely performs arithmetic operations based on input data and outputs the arithmetic operation results to another device located at a remote location.

[0033] The memory 12 includes a volatile storage area (e.g., a working area) that temporarily stores program codes, work memories, etc. when the arithmetic unit 11 executes various programs. Examples of the storage unit include volatile memories such as DRAM (dynamic random access memory) and SRAM (static random access memory), or non-volatile memories such as ROM (Read Only Memory) and flash memory.

[0034] The storage device 13 stores various programs or various data executed by the arithmetic unit 11. The storage device 13 may be one or more non-transitory computer readable media or one or more computer readable storage media. Examples of the storage device 13 include HDD (Hard Disk Drive) and SSD (Solid State Drive).

[0035] The storage device 13 stores the support program 30 and the estimation model 50. The support program 30 describes the content of the support process for estimating support information using the estimation model 50 based on image data (e.g., composite image data) indicating the three-dimensional shape of the biological tissue by the arithmetic unit 11.

[0036] The estimation model 50 includes a neural network 51 and a data set 52 used by the neural network 51. The estimation model 50 is trained to estimate support information based on image data by machine learning using teacher data including image data (e.g., composite image data) indicating the three-dimensional shape of the biological tissue in the oral cavity and support information associated with the image data.

[0037] The neural network 51 may be any algorithm applicable to the neural network 51 in Embodiment 1, such as an autoencoder, a convolutional neural network (CNN), a recurrent neural network (RNN), or a generative adversarial network (GAN). Note that the estimation model 50 may be equipped with other known algorithms such as Bayesian estimation or a support vector machine (SVM), not limited to the neural network 51.

[0038] The dataset 52 includes weight coefficients used in the operations by the neural network 51 and decision thresholds used in the determination during the operations.

[0039] The input interface 14 is an example of the "input unit". The input interface 14 acquires synthetic image data of a biological tissue including at least teeth and gums. The synthetic image data input from the input interface 14 is stored by the memory 12 or the storage device 13 and used when the arithmetic unit 11 estimates support information. Note that the input interface 14 may acquire pre-synthesis IOS data and CT data. For example, the input interface 14 may be communicably connected to a three-dimensional scanner (not shown) and acquire IOS data from the three-dimensional scanner. Also, the input interface 14 may be communicably connected to a CT imaging device (not shown) and acquire CT data from the CT imaging device. In this case, the arithmetic unit 11 generates synthetic image data of a biological tissue including teeth and gums by synthesizing the IOS data and the CT data input from the input interface 14, and estimates support information based on the generated synthetic image data.

[0040] The display interface 15 is an interface for connecting the display 40. The display interface 15 realizes the input and output of data between the support device 1 and the display 40.

[0041] The peripheral device interface 16 is an interface for connecting peripheral devices such as the keyboard 61 and the mouse 62. The peripheral device interface 16 realizes the input and output of data between the support device 1 and the peripheral devices.

[0042] The media reader 17 reads various data stored in the removable disk 20 which is a storage medium, or writes various data to the removable disk 20. For example, the media reader 17 may acquire the support program 30 from the removable disk 20, or may write the support information estimated by the arithmetic unit 11 to the removable disk 20. The removable disk 20 may be one or more non-transitory computer readable media, or may be one or more computer readable storage media. When the arithmetic unit 11 acquires image data (for example, composite image data) from the removable disk 20 via the media reader 17, the media reader 17 can be an example of an "input unit".

[0043] The communication device 18 transmits and receives data to and from an external device via wired communication or wireless communication. For example, the communication device 18 may transmit the support information estimated by the arithmetic unit 11 to an external device (not shown). When the arithmetic unit 11 acquires image data (for example, composite image data) from an external device via the communication device 18, the communication device 18 can be an example of an "input unit".

[0044] [Parameters Estimated in the Support Device] The support device 1 configured as described above is configured to estimate various parameters indicating, as support information, the depth of the periodontal pocket and the like based on the composite image data of the biological tissue including the teeth and the gingiva. With reference to FIGS. 3 and 4, the parameters estimated in the support device 1 according to Embodiment 1 will be described. FIGS. 3 and 4 are diagrams for explaining the parameters estimated in the support device 1 according to Embodiment 1. Note that in FIG. 3, a longitudinal section of the biological tissue including the teeth and the gingiva is shown.

[0045] As shown in FIG. 3, the support device 1 (the arithmetic device 11) sets a predetermined measurement direction for each tooth according to a predetermined standard in the composite image data of the biological tissue including the teeth and the gingiva. For example, while viewing the composite image, the user uses the keyboard 61 and the mouse 62 or the like to specify the tooth axis indicating the inclination of the tooth for each tooth as a predetermined standard to the support device 1. The support device 1 sets the measurement direction along the direction of the tooth axis specified by the user. Note that the user may specify a direction other than the direction of the tooth axis as a predetermined standard to the support device 1. In this case, the support device 1 sets the measurement direction along a direction other than the direction of the tooth axis specified by the user. Note that the support device 1 may set the measurement direction (for example, the tooth axis) by a predetermined mathematical method based on the shape of the biological tissue including the teeth and the gingiva shown in the composite image data.

[0046] When the support device 1 sets the measurement direction, the support device 1 sets at least one predetermined measurement point around each tooth. For example, while viewing the composite image, the user uses the keyboard 61 and the mouse 62 or the like to specify at least one measurement point for each tooth to the support device 1. The measurement point is a point located around the tooth when the crown part included in the tooth is viewed from above. The support device 1 sets at least one measurement point specified by the user. Note that the support device 1 may set at least one measurement point by a predetermined mathematical method based on the shape of the biological tissue including the teeth and the gingiva shown in the composite image data.

[0047] The support device 1 estimates, at each measurement point of each tooth, position information (X, Y, Z) corresponding to each of a plurality of levels indicating the position of the biological tissue existing along the measurement direction, using the estimation model 50. The above-described plurality of levels includes the crown top level, the gingival margin level, the alveolar bone crest level, the gingival boundary level, and the apical level. The crown top level is the level corresponding to the position (for example, height) at the top of the crown portion in the measurement direction. The gingival margin level is the level corresponding to the position (for example, height) of the margin of the gingiva in the measurement direction. The alveolar bone crest level is the level corresponding to the position (for example, height) at the top of the alveolar bone in the measurement direction. The gingival boundary level is the level corresponding to the position (for example, height) of the boundary (the bottom of the periodontal pocket) between the tooth and the gingiva, which is the portion where the gingiva is normally attached to the tooth. The apical level is the level corresponding to the position (for example, height) of the apical portion in the measurement direction. When there is no periodontal pocket or the depth of the periodontal pocket is extremely small, the gingival boundary level and the alveolar bone crest level become the same or substantially the same, and the difference between the gingival boundary level and the alveolar bone crest level is less than a predetermined threshold value.

[0048] When the support device 1 estimates position information (X, Y, Z) corresponding to each of a plurality of levels existing along the measurement direction, it calculates various parameters based on the estimated position information of each level. At this time, the support device 1 calculates different parameters according to whether the gingival boundary level and the alveolar bone crest level are the same or substantially the same.

[0049] Specifically, as shown in FIGS. 3 and 4, when the gingival margin level and the alveolar bone crest level are the same or substantially the same, the support device 1 calculates each of the parameters a to g. Parameter a is a value indicating the distance along the measurement direction between the crown apex level and the gingival margin level. Parameter b is a value indicating the distance along the measurement direction between the crown apex level and the gingival margin level (alveolar bone crest level). Parameter c is a value indicating the distance along the measurement direction between the gingival margin level and the apical level. Parameter d is a value indicating the distance along the measurement direction between the gingival margin level (alveolar bone crest level) and the apical level. Parameter e is a value indicating the distance along the measurement direction between the gingival margin level and the gingival margin level (alveolar bone crest level). Parameter f is the ratio of parameter a to parameter c (a:c). Parameter g is the ratio of parameter b to parameter d (b:d). Note that parameters f and g are also referred to as the crown-root ratio.

[0050] On the other hand, when the gingival margin level and the alveolar bone crest level are not the same or substantially the same, the support device 1 calculates each of the parameters b', d', e', and g'. Parameter b' is a value indicating the distance along the measurement direction between the crown apex level and the gingival margin level. Parameter d' is a value indicating the distance along the measurement direction between the gingival margin level and the apical level. Parameter e' is a value indicating the distance along the measurement direction between the gingival margin level and the gingival margin level. Parameter g' is the ratio of parameter b' to parameter d' (b':d'). Note that parameter g' is also referred to as the crown-root ratio. Thus, when the gingival margin level and the alveolar bone crest level are not the same or substantially the same, the support device 1 omits the calculation of the value indicating the distance along the measurement direction between the crown apex level corresponding to parameter a and the gingival margin level, and the value indicating the distance along the measurement direction between the gingival margin level corresponding to parameter c and the apical level. That is, when the gingival margin level and the alveolar bone crest level are separated due to the pathological condition, the support device 1 may use the gingival margin level instead of the alveolar bone crest level, or may use both the gingival margin level and the alveolar bone crest level.

[0051] As described above, the composite image data is data generated by synthesizing CT data and IOS data. However, the detectable parts from each of the CT data and IOS data are determined.

[0052] FIG. 5 is a diagram for explaining the detection targets in each of the CT data and IOS data. As shown in FIG. 5, the support device 1 can detect the tooth axis, the crown top, the alveolar bone crest, the gingival margin, the root tip, the root bifurcation, the CEJ (Cement Enamel junction), and the measurement point based on the CT data, but cannot detect the gingival edge. The CEJ is the boundary between the cementum located between the tooth root and the periodontal ligament and the enamel covering the tooth emerging from the gingiva, and is also called the cemento-enamel junction. That is, since the CT data can show the tomographic structure or appearance in the hard tissue part (bone, tooth, etc.), it is possible to detect the parts that can be estimated based on the hard tissue part. On the other hand, since the CT data cannot show the soft tissue part (skin, gingiva, etc.) in detail, it is often not possible to sufficiently detect the gingival edge.

[0053] On the other hand, the support device 1 can detect the crown top, the gingival edge, and the measurement point based on the IOS data, but cannot detect the tooth axis, the alveolar bone crest, the gingival margin, the root tip, and the CEJ. That is, since the IOS data can show the surface shape of the living tissue regardless of the hard tissue part and the soft tissue part, it is possible to detect the parts that appear on the surface of the living tissue. On the other hand, since the IOS data cannot show the parts that do not appear on the surface of the living tissue, it is not possible to detect the internal structure of the living tissue such as the alveolar bone crest and the root tip.

[0054] As described above, although both CT data and IOS data have parts that can be detected with high accuracy, the support device 1 can calculate various parameters a~g, b’, d’, e’, g’ as shown in FIG. 4 by using the composite image data obtained by combining the CT data and the IOS data. Further, for each tooth shown by the composite image data, the support device 1 can calculate the above-described various parameters for each measurement point.

[0055] By using the various parameters estimated by the support device 1, the user can diagnose the pathological conditions in periodontal diseases for each measurement point of each tooth. For example, in the case where the gingival margin level and the alveolar bone crest level are the same or substantially the same, it is assumed that the gingiva has not receded and there is a high possibility of not suffering from periodontal disease. In this case, the user can confirm the depth of the periodontal pocket based on the parameter e (the distance between the gingival margin level and the gingival boundary level (alveolar bone crest level)). Further, the user can also confirm the degree of gingival recession based on the parameter f (the ratio of the distance between the crown apex level and the gingival margin level to the distance between the gingival margin level and the root apex level), or the parameter g (the ratio of the distance between the crown apex level and the gingival boundary level (alveolar bone crest level) to the distance between the gingival boundary level (alveolar bone crest level) and the root apex level).

[0056] On the other hand, in the case where the gingival boundary level and the alveolar bone crest level are not the same or substantially the same, it is assumed that the gingiva has receded and there is a high possibility of suffering from periodontal disease. In this case, the user can confirm the depth of the periodontal pocket based on the parameter e’ (the distance between the gingival margin level and the gingival boundary level). Further, the user can also confirm the degree of gingival recession based on the parameter g’ (the ratio of the distance between the crown apex level and the gingival boundary level to the distance between the gingival boundary level and the root apex level).

[0057] In this way, by knowing the various parameters estimated by the support device 1, the user can diagnose the pathological conditions related to periodontal diseases in biological tissues including teeth and gums with low invasiveness and in a short time.

[0058] [Training of the Estimation Model] With reference to FIGS. 6 and 7, the training of the estimation model 50 by machine learning will be described. FIG. 6 is a diagram for explaining an example of machine learning in the learning phase of the estimation model 50 according to Embodiment 1.

[0059] As shown in FIG. 6, in Embodiment 1, in addition to the synthetic image data of biological tissues including teeth and gums, the measurement direction and measurement points are included in the teacher data. For example, during the training of the estimation model 50, machine learning is performed using teacher data including synthetic image data, measurement direction, and measurement points, and the correct data, which is the support information associated with the synthetic image data, the measurement direction, and the measurement points. Further, as the support information, the position information of each level (tooth crown top level, gingival margin level, alveolar bone crest level, gingival boundary level, root apex level) along the measurement direction at each measurement point of each tooth is adopted.

[0060] When the synthetic image data, measurement direction, and measurement points are input, the estimation model 50 estimates the position information (support information) of each level along the measurement direction at each measurement point of each tooth based on the synthetic image data, the measurement direction, and the measurement points by the neural network 51. The estimation model 50 determines whether the estimated position information (support information) of each level matches the position information (support information) of each level, which is the correct data associated with the synthetic image data, the measurement direction, and the measurement points. The estimation model 50 optimizes the dataset 52 by not updating the dataset 52 when the two match, and updating the dataset 52 when the two do not match.

[0061] In this way, the estimation model 50 uses teacher data including the synthetic image data as input data, the measurement direction, and the measurement points, and the position information (support information) of each level as correct answer data. By optimizing the dataset 52, based on the input data, it is trained to accurately estimate the position information of each level (crown top level, gingival margin level, alveolar bone crest level, gingival boundary level, apical level) along the measurement direction at each measurement point of each tooth.

[0062] FIG. 7 is a diagram for explaining an example of the estimation of support information in the utilization phase of the estimation model 50 according to Embodiment 1. As shown in FIG. 7, the estimation model 50 according to Embodiment 1 is trained by machine learning so that when synthetic image data, a measurement direction, and measurement points are input, based on the synthetic image data, the measurement direction, and the measurement points, as support information, the position information of each level (crown top level, gingival margin level, alveolar bone crest level, gingival boundary level, apical level) along the measurement direction at each measurement point of each tooth can be accurately estimated.

[0063] [Data generation process] With reference to FIG. 8, the support process executed by the support device 1 according to Embodiment 1 will be described. FIG. 8 is a flowchart for explaining an example of the support process executed by the support device 1 according to Embodiment 1. Each STEP (hereinafter referred to as "S") shown in FIG. 8 is realized by the arithmetic unit 11 of the support device 1 executing the support program 30.

[0064] As shown in FIG. 8, the support device 1 acquires synthetic image data of a living tissue including teeth and gums (S1). Based on the synthetic image data, the support device 1 sets a predetermined measurement direction for each tooth (S2). For example, the support device 1 sets the tooth axis as the measurement direction for each tooth. At this time, the support device 1 sets the measurement method according to the user's designation. Note that the support device 1 may set the measurement direction by a predetermined mathematical method instead of manually by the user. When the support device 1 automatically sets the measurement direction instead of manually by the user, the user may manually adjust the measurement direction using a keyboard 61, a mouse 62, etc. as necessary.

[0065] Based on the synthetic image data, the support device 1 sets a predetermined measurement point for each tooth (S3). For example, the support device 1 sets, according to the user's designation, a position around a tooth where the depth of a periodontal pocket is generally measured using a probe as the measurement point. Note that the support device 1 may set the measurement point by a predetermined mathematical method instead of manually by the user. When the support device 1 automatically sets the measurement point instead of manually by the user, the user may manually adjust the measurement point using a keyboard 61, a mouse 62, etc. as necessary.

[0066] At each measurement point of each tooth, the support device 1 estimates position information (X, Y, Z) corresponding to each of a plurality of levels existing along the measurement direction using the estimation model 50 (S4). For example, as shown in FIG. 3, the support device 1 estimates the crown top level, the gingival margin level, the alveolar bone crest level, the gingival boundary level, and the apical level using the estimation model 50.

[0067] Based on the estimated levels, the support device 1 determines whether the gingival margin level and the alveolar bone crest level are the same or substantially the same (S5). That is, the support device 1 determines whether the gingival margin level and the alveolar bone crest level are the same or substantially the same without gingival recession, or whether the gingival margin level and the alveolar bone crest level are not the same or substantially the same due to gingival recession. For example, the support device 1 determines whether the difference between the gingival margin level and the alveolar bone crest level is less than a predetermined threshold value.

[0068] When the support device 1 determines that the difference between the gingival margin level and the alveolar bone crest level is less than a predetermined threshold value and the gingival margin level and the alveolar bone crest level are the same or substantially the same (YES in S5), it calculates each of the parameters a to g shown in FIG. 4 (S6). On the other hand, when the support device 1 determines that the difference between the gingival margin level and the alveolar bone crest level is greater than or equal to a predetermined threshold value and the gingival margin level and the alveolar bone crest level are not the same or substantially the same (NO in S5), it calculates each of the parameters b', d', e', g' shown in FIG. 4 (S7).

[0069] The support device 1 displays an image of a living tissue including teeth and gums on the display 40 and adds and displays measurement points to the image (S8). For example, as shown in FIG. 1 and FIG. 9 described later, the support device 1 adds and displays six measurement points around the teeth shown in the image.

[0070] Furthermore, the support device 1 determines the degree of progression of the pathological condition in periodontal disease based on the crown-to-root ratio (parameters f, g, g') calculated in S6 and displays the determination result on the display 40.

[0071] Specifically, the support device 1 determines whether the crown-to-root ratio (parameters f, g, g') calculated in S6 is less than 1 / 2 for one measurement point (S9). Specifically, when the gingival margin level and the alveolar bone crest level are the same or substantially the same, the support device 1 determines whether the parameter f (= a / c) is less than 1 / 2, in other words, whether the distance (c) between the gingival edge level and the root apex level is greater than twice the distance (a) between the crown apex level and the gingival edge level. Alternatively, the support device 1 determines whether the parameter g (= b / d) is less than 1 / 2, in other words, whether the distance (d) between the gingival margin level (alveolar bone crest level) and the root apex level is greater than twice the distance (b) between the crown apex level and the gingival margin level (alveolar bone crest level). Further, when the gingival margin level and the alveolar bone crest level are not the same or substantially the same, the support device 1 determines whether the parameter g' (= b' / d') is less than 1 / 2, in other words, whether the distance (d') between the gingival margin level and the root apex level is greater than twice the distance (b') between the crown apex level and the gingival margin level.

[0072] When the crown-to-root ratio (parameters f, g, g') is less than 1 / 2 (YES in S9), the support device 1 displays the area around the measurement point in a first color (for example, green) to indicate that the gingiva has not regressed and the possibility of suffering from periodontal disease is low (S10). Then, the support device 1 proceeds to the process of S14.

[0073] On the other hand, when the crown-to-root ratio (parameters f, g, g') is greater than 1 / 2 (NO in S9), the support device 1 determines whether the crown-to-root ratio (parameters f, g, g') is less than 1 (S11). Specifically, when the gingival margin level and the alveolar crest level are the same or substantially the same, the support device 1 determines whether the parameter f (= a / c) is less than 1, in other words, whether the distance (c) between the gingival edge level and the apical level is greater than the distance (a) between the crown tip level and the gingival edge level. Alternatively, the support device 1 determines whether the parameter g (= b / d) is less than 1, in other words, whether the distance (d) between the gingival margin level (alveolar crest level) and the apical level is greater than the distance (b) between the crown tip level and the gingival margin level (alveolar crest level). Further, when the gingival margin level and the alveolar crest level are not the same or substantially the same, the support device 1 determines whether the parameter g' (= b' / d') is less than 1, in other words, whether the distance (d') between the gingival margin level and the apical level is greater than the distance (b') between the crown tip level and the gingival margin level.

[0074] When the crown-to-root ratio (parameters f, g, g') is less than 1 (YES in S11), the support device 1 indicates that the gingiva has slightly regressed and that attention should be paid to the possibility of periodontal disease by displaying the area around the measurement point in a second color (for example, orange) (S12). Thereafter, the support device 1 proceeds to the process of S14.

[0075] On the other hand, when the crown-to-root ratio (parameters f, g, g') is greater than 1 (NO in S11), the support device 1 indicates that the gingiva has regressed significantly and that there is a high possibility of periodontal disease by displaying the area around the measurement point in a third color (for example, red) (S13). Thereafter, the support device 1 proceeds to the process of S14.

[0076] In S14, the support device 1 determines whether all measurement points are displayed for at least one tooth shown in the image (S14). If the support device 1 has not displayed all the measurement points (NO in S14), it proceeds to the process of S8. On the other hand, if the support device 1 has displayed all the measurement points (YES in S14), this process ends.

[0077] [Display of Support Information] With reference to FIGS. 9 to 13, an example of the display of support information by the support device 1 according to Embodiment 1 will be described. FIG. 9 is a diagram for explaining a first display example of the support information by the support device 1 according to Embodiment 1.

[0078] As shown in FIG. 9, the support device 1 displays a two - dimensional image of the tooth as seen from above the crown portion on the display 40, and adds and displays six measurement points around the tooth as seen from above the crown portion. When the user moves the cursor using peripheral devices such as the keyboard 61 and the mouse 62 and designates any one of the measurement points, the support device 1 pops up and displays, as support information, the parameters estimated at the designated measurement point together with the tooth number in the vicinity of the designated measurement point.

[0079] For example, the support device 1 displays the tooth number of the tooth specified by the cursor and the depth of the periodontal pocket (parameters e, e') estimated for each of the six measurement points on the tooth. As described in S9 to S13 of FIG. 8, the support device 1 displays the periphery of each measurement point in a color based on the crown-to-root ratio. Further, the support device 1 also displays the depth of the periodontal pocket (parameters e, e') for the measurement point specified by the cursor in a color corresponding to the crown-to-root ratio (f, g, g'). In this example, for the 36th tooth, the values of parameter e (values from 3 mm to 6 mm) estimated for each of the six measurement points are displayed. Further, each measurement point is color-coded in a color based on the crown-to-root ratio (f, g), and the value of parameter e (6 mm) corresponding to the measurement point specified by the cursor is displayed in a color based on the crown-to-root ratio (f, g). That is, the support device 1 displays a color corresponding to the degree of gingival recession indicated by the crown-to-root ratio (f, g). This makes it easier for the user to objectively grasp the periodontal disease situation.

[0080] Note that the support device 1 may display the values of various estimated parameters for the measurement point specified by the cursor, not limited to the depth of the periodontal pocket (parameters e, e'). For example, when the gingival margin level and the alveolar bone crest level are the same or substantially the same, the support device 1 may display any one of parameters a to g. Also, when the gingival margin level and the alveolar bone crest level are not the same or substantially the same, the support device 1 may display any one of parameters b', d', e', g'.

[0081] Note that in the example of FIG. 9, the support device 1 displays the parameter values for only one tooth (the 36th tooth), but may display the parameter values at each measurement point for a plurality of teeth or all teeth.

[0082] In this way, the support device 1 can display, as support information, any one of the parameters a to g estimated using the estimation model 50 or any one of the parameters b', d', e', g' by superimposing them on a designated position of the biological tissue on the display 40. Further, the support device 1 can display, as support information, the depth of the periodontal pocket (parameters e, e') estimated by the support device 1 in a color corresponding to the crown-to-root ratio (f, g, g') on the display 40. The values of the color-coded respective parameters displayed on such a display 40 can be information indicating the degree of progression of the pathological condition in the biological tissue in the oral cavity such as periodontal disease.

[0083] FIG. 10 is a diagram for explaining a second display example of the support information by the support device 1 according to the first embodiment. As shown in FIG. 10, the support device 1 may create a chart summarizing the depth of the periodontal pocket in each tooth and display the created chart on the display 40.

[0084] For example, the support device 1 displays, as support information, the depth of the periodontal pocket (parameters e, e') estimated for each of the six measurement points in each tooth. In the example of FIG. 10, the support device 1 displays the depth of the periodontal pocket (parameters e, e') for only one tooth (tooth No. 36), but may display the depth of the periodontal pocket (parameters e, e') at each measurement point for a plurality of teeth or all teeth.

[0085] FIG. 11 is a diagram for explaining a third display example of support information by the support device 1 according to Embodiment 1. As shown in FIG. 11, the support device 1 displays a two-dimensional image simulating a dental arch on the display 40, and may display the depth of the periodontal pocket (parameters e, e') as support information for each measurement point of each aligned tooth. For example, in FIG. 11, images of the front side and the back side of each tooth included in each of the upper dental arch and the lower dental arch are displayed on the display 40, and points are plotted at the positions of the gingival boundary levels when the gingival edge level of each tooth is set to "0". In the example of FIG. 11, the support device 1 displays only the depth of the periodontal pocket (parameters e, e') of each tooth as support information, but any of the parameters a to g estimated using the estimation model 50 or any of the parameters b', d', e', g' may be displayed.

[0086] FIG. 12 is a diagram for explaining a fourth display example of support information by the support device according to Embodiment 1. As shown in FIG. 12, the support device 1 displays a three-dimensional image showing a dental arch or the oral cavity on the display 40, and may pop-up display the parameters estimated at the specified measurement point together with the tooth number in the vicinity of the measurement point specified by the user as support information.

[0087] For example, the support device 1 may display the tooth number and the depth of the periodontal pocket (parameters e, e') estimated for each of the 6 measurement points on the tooth specified by the cursor. Further, the support device 1 may display the depth of the periodontal pocket (parameters e, e') in a color corresponding to the crown-root ratio (f, g, g') for the portion of the tooth specified by the cursor. In this example, the value of parameter e (values from 3 mm to 6 mm) estimated for each of the 6 measurement points on tooth No. 36 is displayed. Furthermore, the value of parameter e (6 mm) corresponding to the measurement point specified by the cursor is displayed in a color based on the crown-root ratio (f, g).

[0088] The support device 1 may color-code and display the gingiva near each tooth with a color based on the crown-to-root ratio (f, g, g'). For example, the support device 1 may show the user the degree of the depth of the periodontal pocket (parameters e, e') by highlighting the color of the gingiva using a heat map or the like according to the crown-to-root ratio. Thereby, the user can objectively grasp the degree of the depth of the periodontal pocket (parameters e, e') for each part of the dentition.

[0089] Furthermore, the support device 1 may calculate and display any one of parameters a to g or any one of parameters b', d', e', g' for continuous portions (for example, measurement points exceeding 6 points) around each tooth, or may calculate and display an average value, a deviation, or the like for the crown-to-root ratio (f, g, g') for each individual tooth. The support device 1 may highlight the color of the gingiva so that the user can objectively grasp the depth of the periodontal pocket.

[0090] Note that the support device 1 may display the values of various estimated parameters not only for the depth of the periodontal pocket (parameters e, e') but also for the measurement points designated by the cursor. For example, when the gingival boundary level and the alveolar bone crest level are the same or substantially the same, the support device 1 may display any one of parameters a to g. Also, when the gingival boundary level and the alveolar bone crest level are not the same or substantially the same, the support device 1 may display any one of parameters b', d', e', g'.

[0091] FIG. 13 is a diagram for explaining a fifth display example of support information by the support device according to Embodiment 1. As shown in FIG. 13, the support device 1 displays a three-dimensional composite image of a biological tissue including teeth and gingiva on the display 40, and near the measurement point designated by the user, together with the tooth number, as support information, may pop up and display the parameters estimated at the designated measurement point.

[0092] For example, the support device 1 may display the tooth number of the tooth specified by the cursor and the depth of the periodontal pocket (parameters e, e') estimated for each of the six measurement points at the tooth. Further, the support device 1 may display the depth of the periodontal pocket (parameters e, e') in a color corresponding to the crown-to-root ratio (f, g, g') for the portion of the tooth specified by the cursor. In this example, at the tooth number 46, the value of parameter e (a value between 3 mm and 6 mm) estimated for each of the six measurement points is displayed. Also, the support device 1 may display the value of parameter e (6 mm) corresponding to the measurement point specified by the cursor in a color based on the crown-to-root ratio (f, g).

[0093] Furthermore, the support device 1 may calculate and display any one of parameters a to g or any one of parameters b', d', e', g' for continuous portions around each tooth (for example, measurement points exceeding six points), and for the crown-to-root ratio (f, g, g'), an average value or deviation or the like may be calculated and displayed for each individual tooth. The support device 1 may highlight the color of the gingiva so that the user can objectively grasp the depth of the periodontal pocket.

[0094] Note that the support device 1 may display the values of various estimated parameters for the measurement point specified by the cursor, not limited to the depth of the periodontal pocket (parameters e, e'). For example, when the gingival margin level and the alveolar bone crest level are the same or substantially the same, the support device 1 may display any one of parameters a to g. Also, when the gingival margin level and the alveolar bone crest level are not the same or substantially the same, the support device 1 may display any one of parameters b', d', e', g'.

[0095] As described above, based on the image data (synthetic image data) of the synthetic image generated based on the IOS data and the CT data, for each tooth, the support device 1 estimates various parameters of a to g or b', d', e', g', and presents the values of these parameters to the user. As a result, the user can non-invasively check the depth of the periodontal pocket without inserting a probe into the patient's periodontal pocket, and can accurately check the progression of periodontal disease without relying on their own skills.

[0096] Furthermore, as shown in FIGS. 9 to 13, the support device 1 displays a two-dimensional or three-dimensional image including teeth on the display 40, and also displays the values of various parameters such as the depth of the periodontal pocket (parameters e, e') at the designated measurement points, or color-codes the gums or the values of the parameters based on the crown-to-root ratio (f, g, g'). As a result, the user can easily grasp the degree of progression of the pathological condition of periodontal disease, which makes it easier to explain to the patient and also easier for the patient to understand.

[0097] [Modification Example of Embodiment 1] A modification example of the support device 1 according to Embodiment 1 will be described with reference to FIGS. 14 to 19. In the modification example of the support device 1 according to Embodiment 1, only the parts different from the support device 1 according to Embodiment 1 will be described, and the same parts as the support device 1 according to Embodiment 1 will be given the same reference numerals and their descriptions will not be repeated.

[0098] The first modification example will be described. FIG. 14 is a diagram for explaining an example of machine learning in the learning phase of the estimation model 50 according to the first modification example. As shown in FIG. 14, in the first modification example, the measurement direction and the measurement point are not included in the teacher data. Also, as the correct data, i.e., the support information, the position information of each level (crown top level, gum margin level, alveolar bone crest level, gum boundary level, root apex level) along the measurement direction at each measurement point of each tooth is adopted.

[0099] When the synthetic image data is input, the estimation model 50 estimates, based on the synthetic image data, the position information (support information) at each level along the measurement direction at each measurement point of each tooth by means of the neural network 51. At this time, although the measurement direction and the measurement point are not input, the estimation model 50 estimates, by its own estimation, the position information (support information) at each level along the measurement direction at each measurement point of each tooth. The estimation model 50 determines whether or not the position information (support information) at each level along the measurement direction at each measurement point of each tooth estimated by the estimation model 50 matches the position information (support information) at each level along the measurement direction at each measurement point of each tooth, which is the correct answer data associated with the synthetic image data. When the two match, the estimation model 50 does not update the data set 52, while when the two do not match, the estimation model 50 updates the data set 52 to optimize the data set 52.

[0100] In this way, the estimation model 50 uses the teacher data including the synthetic image data as the input data and the position information (support information) at each level as the correct answer data, and through the optimization of the data set 52, it is trained to accurately estimate the position information at each level (crown top level, gingival margin level, alveolar bone crest level, gingival boundary level, root apex level) along the measurement direction at each measurement point of each tooth based on the input data. That is, the estimation model 50 is trained to be able to accurately estimate the position information at each level along the measurement direction at each measurement point of each tooth based on the input synthetic image data, even if the measurement direction and the measurement point are not input, by performing machine learning on the measurement direction and the measurement point that are not input.

[0101] FIG. 15 is a diagram for explaining an example of estimation of support information in the utilization phase of the estimation model 50 according to the first modification. As shown in FIG. 15, the estimation model 50 according to the first modification is trained by machine learning so that when synthetic image data is input, based on the synthetic image data, as support information, for each measurement point of each tooth, position information of each level (crown top level, gingival margin level, alveolar bone crest level, gingival boundary level, apical level) along the measurement direction can be estimated. Thus, according to the estimation model 50 according to the first modification, without setting the measurement direction and measurement point according to the user's designation as in S2 and S3 of FIG. 8, the measurement direction and measurement point can be automatically estimated, and the position information of each level along the measurement direction at each measurement point of each tooth can be accurately estimated.

[0102] Note that in the learning phase, either one of the measurement direction and the measurement point may be included in the input data together with the synthetic image data. In this case, in the utilization phase, the estimation model 50 can accurately estimate the position information of each level along the measurement direction at each measurement point of each tooth based on the input data including either one of the measurement direction and the measurement point and the synthetic image data.

[0103] The second modification will be described. FIG. 16 is a diagram for explaining an example of machine learning in the learning phase of the estimation model 50 according to the second modification. As shown in FIG. 16, in the second modification, as support information, each parameter (a to g or b', d', e', g') at each measurement point of each tooth is adopted. For example, when the gingival boundary level and the alveolar bone crest level are the same or substantially the same in the tooth to be estimated, each parameter from a to g at each measurement point of each tooth is adopted as support information. Also, when the gingival boundary level and the alveolar bone crest level are not the same or substantially the same in the tooth to be estimated, each parameter of b', d', e', g' at each measurement point of each tooth is adopted as support information.

[0104] During the training of the estimation model 50, machine learning is performed using teacher data including synthetic image data, a measurement direction, and a measurement point as input data, and support information (a to g or b', d', e', g') associated with the synthetic image data, the measurement direction, and the measurement point as correct answer data. Note that either one of the measurement direction and the measurement point may be included in the input data together with the synthetic image data.

[0105] When synthetic image data, a measurement direction, and a measurement point are input, the estimation model 50 estimates each parameter (a to g or b', d', e', g') at each measurement point of each tooth based on the image data, the measurement direction, and the measurement point by the neural network 51. The estimation model 50 determines whether or not the estimated each parameter (a to g or b', d', e', g') (support information) matches each parameter (a to g or b', d', e', g') at each measurement point of each tooth, which is the correct answer data associated with the image data, the measurement direction, and the measurement point. The estimation model 50 optimizes the dataset 52 by updating the dataset 52 when they do not match, while not updating the dataset 52 when they match.

[0106] In this way, the estimation model 50 uses teacher data including synthetic image data, a measurement direction, and a measurement point as input data, and each parameter (a to g or b', d', e', g') (support information) at each measurement point of each tooth as correct answer data, and is trained to be able to accurately estimate each parameter (a to g or b', d', e', g') at each measurement point of each tooth based on the input data by optimizing the dataset 52.

[0107] FIG. 17 is a diagram for explaining an example of estimation of support information in the utilization phase of the estimation model 50 according to the second modification. As shown in FIG. 17, the estimation model 50 according to the second modification is trained by machine learning so that when synthetic image data, a measurement direction, and a measurement point are input, based on the image data, the measurement direction, and the measurement point, as support information, each parameter (a to g or b', d', e', g') at each measurement point of each tooth can be accurately estimated.

[0108] In the learning phase, at least one of a plurality of parameters (a to g or b', d', e', g') may be included in the support information which is correct data. In this case, in the utilization phase, the estimation model 50 estimates, as support information, at least one of a plurality of parameters (a to g or b', d', e', g') at each measurement point of each tooth based on input data including synthetic image data, a measurement direction, and a measurement point.

[0109] The third modification will be described. FIG. 18 is a diagram for explaining an example of machine learning in the learning phase of the estimation model 50 according to the third modification. As shown in FIG. 18, in the third modification, the type of pathological condition and the degree of progression of the pathological condition in the biological tissue in the oral cavity are adopted as support information.

[0110] During the training of the estimation model 50, machine learning is performed using teacher data including synthetic image data, a measurement direction, and a measurement point as input data, and support information (type of pathological condition and degree of progression of the pathological condition) associated with the image data, the measurement direction, and the measurement point as correct data. Note that either one of the measurement direction and the measurement point may be included in the input data together with the image data.

[0111] When the synthetic image data, measurement direction, and measurement point are input, the estimation model 50 estimates the type of pathological condition and the degree of progression of the pathological condition in the living tissue in the oral cavity based on the image data, the measurement direction, and the measurement point by means of the neural network 51. The estimation model 50 determines whether or not the estimated type of pathological condition and the degree of progression of the pathological condition (support information) match the type of pathological condition and the degree of progression of the pathological condition in the living tissue in the oral cavity, which is the correct data associated with the image data, the measurement direction, and the measurement point. When they match, the estimation model 50 does not update the data set 52, while when they do not match, the estimation model 50 updates the data set 52 to optimize the data set 52.

[0112] In this way, the estimation model 50 uses the teacher data including the synthetic image data, measurement direction, and measurement point as the input data and the type of pathological condition and the degree of progression of the pathological condition (support information) in the living tissue in the oral cavity as the correct data. By optimizing the data set 52, the estimation model 50 is trained to accurately estimate the type of pathological condition and the degree of progression of the pathological condition in the living tissue in the oral cavity based on the input data.

[0113] FIG. 19 is a diagram for explaining an example of the estimation of support information in the utilization phase of the estimation model 50 according to the third modification. As shown in FIG. 19, the estimation model 50 according to the third modification is trained by machine learning so that when the synthetic image data, measurement direction, and measurement point are input, based on the image data, the measurement direction, and the measurement point, it can accurately estimate the type of pathological condition and the degree of progression of the pathological condition in the living tissue in the oral cavity as support information.

[0114] In the learning phase, at least one of the type of the pathological condition and the degree of progression of the pathological condition may be included in the support information that is the correct data. In this case, in the utilization phase, the estimation model 50 estimates at least one of the type of the pathological condition and the degree of progression of the pathological condition as support information based on the input data including the synthetic image data, the measurement direction, and the measurement point.

[0115] As described above, the support device 1 may estimate, as diagnostic information, at least one of a plurality of levels (crown top level, gingival margin level, alveolar bone crest level, gingival boundary level, root apex level) using the estimation model 50, or may estimate at least one of the parameters (a to g or b', d', e', g') without estimating the level, or may estimate at least one of the type of the pathological condition and the degree of progression of the pathological condition in periodontal disease without estimating the level and the parameters.

[0116] In addition, for a tooth affected by periodontal disease, as the gingiva recedes, more of the surface portion of the tooth appears than in a tooth not affected by periodontal disease. That is, the color of the surface of the living tissue including the tooth and the gingiva differs depending on whether the tooth is affected by periodontal disease or not. Therefore, the support device 1 may perform machine learning by adopting, as input data of teacher data, the color information indicating the color of the surface of the living tissue included in the three-dimensional data acquired by the three-dimensional scanner together with the synthetic image data. In this case, in addition to the synthetic image data, the support device 1 can accurately estimate the degree of progression of the pathological condition in periodontal disease based on the color of the surface of the living tissue.

[0117] <Embodiment 2> The support device 1 according to Embodiment 2 will be described with reference to FIGS. 20 to 22. In the support device 1 according to Embodiment 2, only the parts different from the support device 1 according to Embodiment 1 will be described, and the same parts as the support device 1 according to Embodiment 1 will be denoted by the same reference numerals and their description will not be repeated.

[0118] FIG. 20 is a diagram for explaining an example of machine learning in the learning phase of the estimation model 50 according to Embodiment 2. As shown in FIG. 20, in Embodiment 2, in addition to the synthetic image data, at least one of information related to the gender, age, and bone density of a patient having a biological tissue represented by the synthetic image data is included in the teacher data. The information related to the bone density includes at least one of the CT value (the black-and-white image density value in the CT image) of the jawbone around an individual tooth and the trabecular bone structure index (TBS) of the jawbone around an individual tooth. Further, as the support information, the degree of progression of the pathological condition in the oral biological tissue is adopted.

[0119] Here, depending on information (CT value, trabecular bone structure index) related to the gender, age, and bone density of the patient, the degree of progression of the pathological condition in periodontal disease tends to be different. For example, it is said that women are more likely to suffer from periodontal disease than men. Also, aging makes it easier to suffer from periodontal disease. Furthermore, the lower the bone density, the more likely it is to suffer from periodontal disease. Therefore, if at least one of information (CT value, trabecular bone structure index) related to the gender, age, and bone density of the patient is adopted as the teacher data for the input data of the estimation model 50, the estimation model 50 can estimate the degree of progression of the pathological condition in periodontal disease.

[0120] For example, during the training of the estimation model 50, machine learning is performed using teacher data including synthetic image data, information related to gender, age, and bone density (CT value, trabecular bone structure index) as input data, and support information (degree of progression of the pathological condition) associated with the image data, the gender, the age, and the information related to the bone density as the correct data. Note that either one of the measurement direction and the measurement point may be included in the input data.

[0121] When synthetic image data, gender, age, and information related to bone density (CT value, trabecular bone structure index) are input into the estimation model 50, the neural network 51 estimates the degree of progression of the pathological condition in the living tissue in the oral cavity based on the image data, the gender, the age, and the information related to the bone density. The estimation model 50 determines whether the estimated degree of progression of the pathological condition (support information) matches the degree of progression of the pathological condition in the living tissue in the oral cavity, which is the correct data associated with the image data, the gender, the age, and the information related to the bone density. When the two match, the estimation model 50 does not update the data set 52, while when they do not match, the estimation model 50 optimizes the data set 52 by updating the data set 52.

[0122] In this way, the estimation model 50 uses the teacher data including the synthetic image data, gender, age, and information related to bone density (CT value, trabecular bone structure index) as input data and the degree of progression of the pathological condition (support information) in the living tissue in the oral cavity as the correct data. By optimizing the data set 52, the estimation model 50 is trained to accurately estimate the degree of progression of the pathological condition in the living tissue in the oral cavity based on the input data.

[0123] FIG. 21 is a diagram for explaining an example of the estimation of support information in the utilization phase of the estimation model 50 according to Embodiment 2. As shown in FIG. 21, the estimation model 50 according to Embodiment 2 is trained by machine learning so that when synthetic image data, gender, age, and information related to bone density (CT value, trabecular bone structure index) are input, the degree of progression of the pathological condition in the living tissue in the oral cavity can be accurately estimated based on the image data, the gender, the age, and the information related to the bone density.

[0124] In addition, in the learning phase, at least one of the information related to gender, age, and bone density (CT value, trabecular bone structure index) may be included in the input data in addition to the synthetic image data. In this case, in the utilization phase, the estimation model 50 estimates the degree of progression of the disease state as support information based on the synthetic image data and at least one of the information related to gender, age, and bone density (CT value, trabecular bone structure index).

[0125] In this way, the support device 1 can accurately estimate the degree of progression of the disease state in periodontal disease without relying on the experience of the operator and while reducing the burden on the operator by using at least one of the information related to the patient's gender, age, and bone density (CT value, trabecular bone structure index) that can affect the degree of progression of the disease state in periodontal disease as the input data of the teacher data and performing machine learning. Also, if the user estimates and records the degree of progression of the disease state in the patient's periodontal disease using the support device 1 every year in the national dental health check that will be introduced in the future, the degree of progression of the disease state in the future can also be predicted based on the record.

[0126] Furthermore, the estimation model 50 may be machine-learned and trained using the teacher data with the record of the annual degree of progression of the disease state as described above as the input data and the degree of progression of the disease state as the correct answer data. In this case, when the current degree of progression of the disease state is input, the estimation model 50 can also estimate the future degree of progression of the disease state based on the current degree of progression of the disease state.

[0127] FIG. 22 is a diagram for explaining an example of output of support information by the support device 1 according to the second embodiment. As shown in FIG. 22, the support device 1 may score the degree of progression of the disease state in periodontal disease estimated using the estimation model 50 based on a predetermined determination criterion and display the calculated score on the display 40.

[0128] For example, as shown in Fig. 22(A), the support device 1 may estimate the degree of progression of the pathological condition in periodontal disease for the entire dentition, and calculate and display a score corresponding to the degree of progression of the pathological condition for each of "normal", "progressive (requiring observation)", and "unfavorable (exceeding the tolerance)" based on a predetermined determination criterion. As shown in Fig. 22(B), the support device 1 may estimate the degree of progression of the pathological condition in periodontal disease for each of the upper left dentition, the upper right dentition, the lower left dentition, and the lower right dentition, and calculate and display a score corresponding to the degree of progression of the pathological condition for each of "normal", "progressive (requiring observation)", and "unfavorable (exceeding the tolerance)" based on a predetermined determination criterion. As shown in Fig. 22(C), the support device 1 may estimate the degree of progression of the pathological condition in periodontal disease for each tooth included in each of the upper left dentition, the upper right dentition, the lower left dentition, and the lower right dentition, and calculate and display a score corresponding to the degree of progression of the pathological condition for each of "normal", "progressive (requiring observation)", and "unfavorable (exceeding the tolerance)" based on a predetermined determination criterion.

[0129] In addition, the support device 1 may display a score corresponding to the degree of progression of the pathological condition in the image of the tooth as shown in Figs. 9 to 13. For example, the support device 1 may highlight the color of the gingiva using a heat map or the like according to the degree of progression of the pathological condition in the image of the living tissue including the tooth and the gingiva as shown in Fig. 12 or Fig. 13. The support device 1 may show the user simulation information predicting the future change of the degree of progression of the pathological condition (support information) by displaying the image showing such a degree of progression of the pathological condition in time series for each future elapsed time (for example, one year later, two years later, etc.).

[0130] <Embodiment 3> The support device 1 according to Embodiment 3 will be described with reference to Figs. 23 and 24. In the support device 1 according to Embodiment 3, only the parts different from the support device 1 according to Embodiment 1 will be described, and the same parts as the support device 1 according to Embodiment 1 will be denoted by the same reference numerals and their description will not be repeated.

[0131] FIG. 23 and FIG. 24 are diagrams for explaining the parameters estimated in the support device 1 according to Embodiment 3. In FIG. 23, a longitudinal section of a biological tissue including teeth and gums is shown.

[0132] As shown in FIG. 23, in the support device 1 according to Embodiment 3, at each measurement point of each tooth, using the estimation model 50, as a plurality of levels indicating the positions of biological tissues existing along the measurement direction, position information (X, Y, Z) corresponding to each of the CEJ level, the gingival margin level, the furcation level, the deepest bone defect level, and the apical level is estimated. The CEJ level is a level corresponding to the position (for example, height) of the cementum located between the tooth root and the periodontal ligament and the enamel covering the tooth emerging from the gum (cemento-enamel junction). The furcation level is a level corresponding to the position (for example, height) of the furcation. The deepest bone defect level is a level corresponding to the position (for example, height) of the deepest part of the defect of the alveolar bone when the furcation is diseased and a space is formed between the furcation and the alveolar bone.

[0133] By estimating the CEJ level, the gingival margin level, the furcation level, the deepest bone defect level, and the apical level at each measurement point of each tooth, the support device 1 can derive the degree of progression of the pathological condition in the diseased furcation. Specifically, as shown in FIGS. 23 and 24, the support device 1 calculates each parameter of parameters h to l. Parameter h is a value indicating the distance along the measurement direction between the CEJ level and the furcation level. Parameter i is a value indicating the distance along the measurement direction between the CEJ level and the deepest bone defect level. Parameter j is a value indicating the distance along the measurement direction between the CEJ level and the gingival margin level. Parameter k is a value indicating the distance along the measurement direction between the furcation level and the deepest bone defect level. Parameter l is a value indicating the distance along the measurement direction between the deepest bone defect level and the apical level.

[0134] By using various parameters estimated by the support device 1, the user can diagnose the pathological conditions at each measurement point of each tooth regarding the diseased furcation area. For example, based on at least one of the parameters h to l estimated by the support device 1, the user can diagnose the pathological conditions at the diseased furcation area by using known Glickman's furcation lesion classification, Lindhe's furcation lesion classification, Tarnow & Fletcher's classification, or the like.

[0135] In this way, by knowing the various parameters estimated by the support device 1, the user can diagnose the pathological conditions related to furcation lesions in biological tissues including teeth and gums with low invasiveness and in a short time.

[0136] Note that the support device 1 may estimate at least one of a plurality of levels (CEJ level, gingival margin level, furcation level, deepest bone defect level, apical level) as diagnostic information by using the estimation model 50, or may estimate at least one of the parameters h to l without estimating the level, or may estimate the degree of progression of the pathological conditions at the diseased furcation area without estimating the level and the parameters.

[0137] Note that the support device 1 according to each of the above-described Embodiment 1 and Embodiment 2 may include the configurations and functions of each other alone or in combination. Furthermore, the support device 1 according to each of Embodiment 1 and Embodiment 2 may include the configurations and functions of the above-described modified examples alone or in combination.

[0138] It should be considered that all aspects of the embodiments disclosed this time are illustrative and not restrictive. The scope of the present disclosure is shown by the claims rather than the above description, and it is intended that all changes within the meaning and scope equivalent to the claims are included. Note that the configurations illustrated in this embodiment and the configurations illustrated in the modified examples can be combined as appropriate.

Explanation of Reference Numerals

[0139] 1 Support device, 11 Computing device, 12 Memory, 13 Storage device, 14 Input interface, 15 Display interface, 16 Peripheral device interface, 17 Media reader, 18 Communication device, 20 Removable disk, 30 Support program, 40 Display, 50 Estimation model, 51 Neural network, 52 Dataset, 61 Keyboard, 62 Mouse.

Claims

1. An assistance device for assisting in the diagnosis of a pathological condition in a living tissue in the oral cavity including at least teeth and gums, an input unit into which input data including image data showing a three-dimensional shape of the living tissue, measurement points of the living tissue designated for diagnosing the pathological condition, and a measurement direction at the measurement points is input, and an arithmetic unit that derives the assistance information using the input data and an estimation model trained by machine learning so as to derive assistance information for assisting in the diagnosis of the pathological condition in the living tissue based on the input data. The arithmetic unit, derives position information corresponding to each of a plurality of levels indicating positions of the living tissue existing along the measurement direction at the measurement points, calculates a distance between the plurality of levels based on the position information corresponding to each of the plurality of levels, and calculates at least one of information indicating the type of the pathological condition and information indicating the degree of progression of the pathological condition as the assistance information based on the distance between the plurality of levels. An assistance device.

2. The assistance device according to claim 1, wherein the estimation model is trained by the machine learning using teacher data including the input data and the assistance information associated with the input data.

3. The assistance device according to claim 1, wherein the measurement points are points located around the teeth when the crown portions included in the teeth are viewed from above.

4. The assistance device according to claim 1, wherein the measurement direction is a direction along the tooth axis of the teeth.

5. The plurality of levels indicate any one of the top of the crown portion, the edge of the gum, the top of the alveolar bone, the boundary between the tooth and the gum, the apical portion, the root bifurcation portion, the boundary between the cementum and the enamel of the tooth, and the deepest part of the defective portion of the alveolar bone defective at the root bifurcation portion in the living tissue. The assistance device according to claim 1.

6. The assistance device according to claim 1, wherein the input data includes at least one of information related to the gender, age, and bone density of a patient having the living tissue.

7. The support device according to claim 1, wherein the image data is generated based on optical scanner data captured by an optical scanner including position information of each point of a point cloud indicating the surface of the living tissue, and CT (Computed Tomography) data obtained by subjecting the living tissue to computed tomography.

8. The support device according to claim 7, wherein the optical scanner data includes information indicating the color of the surface of the living tissue.

9. The support device according to claim 1, wherein the calculation unit causes the support information to be displayed on the display by superimposing it on a designated position of the living tissue.

10. The support device according to claim 9, wherein the calculation unit causes the support information to be displayed on the display in a color corresponding to the support information.

11. The support device according to claim 9, wherein the calculation unit causes simulation information predicting a future change of the support information to be displayed on the display based on the support information.

12. A support method for assisting in the diagnosis of a pathological condition in a living tissue in the oral cavity including at least teeth and gums by a computer, wherein the support method includes, as a process executed by the computer, a step of acquiring input data including image data indicating a three-dimensional shape of the living tissue, measurement points of the living tissue designated for diagnosing the pathological condition, and a measurement direction at the measurement points; a step of deriving the support information using the input data and an estimation model trained by machine learning to derive support information for assisting in the diagnosis of the pathological condition in the living tissue based on the input data. The step of deriving includes a step of deriving position information corresponding to each of a plurality of levels indicating positions of the living tissue existing along the measurement direction at the measurement points; a step of calculating a distance between the plurality of levels based on the position information corresponding to each of the plurality of levels; a step of calculating, as the support information, at least one of information indicating the type of the pathological condition and information indicating the degree of progression of the pathological condition based on the distance between the plurality of levels.

13. A support program for assisting in the diagnosis of a pathological condition in a living tissue in the oral cavity including at least teeth and gums by a computer, wherein the computer is caused to Obtaining input data including image data showing the three-dimensional shape of the living tissue, measurement points of the living tissue specified for diagnosing the pathological condition, and measurement directions at the measurement points; Deriving the support information using the input data and an estimation model trained by machine learning to derive support information for assisting in diagnosing the pathological condition in the living tissue based on the input data; The step of deriving includes: Deriving position information corresponding to each of a plurality of levels indicating positions of the living tissue existing along the measurement direction at the measurement point; Calculating distances between the plurality of levels based on the position information corresponding to each of the plurality of levels; A support program including calculating, as the support information, at least one of information indicating the type of the pathological condition and information indicating the degree of progression of the pathological condition based on the distances between the plurality of levels.

Images