JP2024022277A - Periodontal pocket measurement apparatus and periodontal pocket measurement method - Google Patents

Abstract

To provide a periodontal pocket measurement apparatus and a periodontal pocket measurement method which can correctly and easily measure a depth of a periodontal pocket.SOLUTION: A periodontal pocket measurement apparatus 1 comprises: a scanner 111 which acquires an oral cavity shape by imaging an oral cavity; an air blower 113 which jets compressed air to an imaging object when the scanner performs the imaging; and an information processing device 13. The scanner acquires point cloud data indicating the oral cavity shape while the air blower jets the compressed air to the imaging object. The information processing device comprises: a sensing data acquisition unit 131 which receives the point cloud data from the scanner; a model generation unit 133 which generates a model on the basis of the point cloud data; a recognition unit 135 which identifies a gum, a tooth root and a gingiva on the basis of a feature of the model; and a measurement unit 137 which measures a depth of a periodontal pocket on the basis of the identification result of the gum, the tooth root and the gingiva.SELECTED DRAWING: Figure 1

Description

The present invention relates to a periodontal pocket measuring device and a periodontal pocket measuring method, and specifically relates to a technique for accurately and simply measuring the depth of a periodontal pocket.

The depth of periodontal pockets is used as an indicator of the degree of progress of periodontal disease. Periodontal pocket depth refers to the depth of the groove between the teeth and the gums, or the gingival sulcus. The depth of periodontal pockets is approximately 1 mm to 3 mm in a healthy state, but may increase from 3 mm to 6 mm or more as periodontal disease progresses.

Conventionally, a needle-shaped instrument with a scale called a probe has been used to measure the depth of periodontal pockets. The probe is inserted into the gingival sulcus, and the practitioner visually confirms the depth when it stops. Do this at several points on the gums for each tooth. Measuring the depth of periodontal pockets through such a process not only requires a lot of effort and time, but also has the problem of unavoidable variations in measurement results depending on the operator. For example, there is a large degree of variation in the use of evidence to trace the progress of periodontal disease treatment in a patient, and a more stable and reliable measurement method was desired.

On the other hand, with the advancement of digitalization (digital dentistry) in the dental industry, equipment that can perform treatment more easily and with higher precision than before has appeared. For example, a device called an Intra Oral Scanner (IOS) can photograph the inside of a patient's oral cavity using a reader called a wand and output the shape of the oral cavity as digital data (model). Typically, a wand emits light toward an object (teeth, gums, etc.), and a sensor detects the reflected light, thereby acquiring point cloud data indicating the three-dimensional shape of the object.

However, since IOS can only image the visible region within the oral cavity, it is usually not possible to model the invisible region below the gingival sulcus. Therefore, IOS has been considered unsuitable for measuring the depth of periodontal pockets.

Related technologies include Patent Documents 1 and 2. Patent Documents 1 and 2 disclose intraoral scanners that acquire the shape of an abutment tooth located below the gingival margin by tracing the tooth surface with a stylus-like measuring jig.

Japanese Patent Application Publication No. 2018-047299 Special table 2016-508754 publication JP2017-213060A

Ryo Onodera et al., “Correction method for colored point clouds using laser reflection intensity”, [online], University of Electronics and Communication, [searched on July 26, 2020], Internet <URL: http://www. ddm. mi. uec. ac. jp/papers/smt2015s_onodera. pdf>

The present invention has been made to solve these problems, and provides a periodontal pocket measuring device and a periodontal pocket measuring method that can accurately and easily measure the depth of periodontal pockets. The purpose is to

A periodontal pocket measuring device according to an embodiment of the present invention includes a scanner that photographs the inside of the oral cavity to obtain the shape of the oral cavity, and an air scanner that injects compressed air to the object to be photographed when the scanner performs the photographing. The scanner includes a blower and an information processing device, and the scanner acquires point cloud data indicating the intraoral shape while the air blower injects the compressed air to the imaging target, and performs the information processing. The device includes a sensing data acquisition unit that receives the point cloud data from the scanner, a model generation unit that generates a model based on the point cloud data, and identifies gums, tooth roots, and gingiva based on the characteristics of the model. and a measuring section that measures the depth of the periodontal pocket based on the identification results of the gums, tooth roots, and gingiva.
The periodontal pocket measuring device according to one embodiment further includes a determination unit that determines the degree of progress of periodontal disease based on at least the depth of the periodontal pocket.
In the periodontal pocket measuring device according to one embodiment, the scanner continuously acquires a plurality of the point cloud data, and the recognition unit recognizes characteristics of changes in position of the point cloud data over time. Identify teeth and gums based on.
In the periodontal pocket measuring device according to one embodiment, the scanner acquires the point cloud data having color information, and the recognition unit detects the gums, Identify the tooth root and gingiva.
In the periodontal pocket measuring method according to an embodiment of the present invention, a scanner images the inside of the oral cavity while an air blower injects the compressed air to the object to be imaged, and the intraoral shape is obtained. a sensing data acquisition step of receiving point cloud data indicating the point cloud; a model generation step of generating a model based on the point cloud data; a recognition step of identifying gums, tooth roots, and gingiva based on the characteristics of the model; and a measuring step of measuring the depth of the periodontal pocket based on the identification results of the gums, tooth roots, and gingiva.
The periodontal pocket measuring method according to one embodiment further includes a determination step of determining the degree of progress of periodontal disease based on at least the depth of the periodontal pocket.
A program according to an embodiment of the present invention causes a computer to execute the above method.

According to the present invention, it is possible to provide a periodontal pocket measuring device and a periodontal pocket measuring method that can accurately and easily measure the depth of a periodontal pocket.

1 is a block diagram showing the configuration of a periodontal pocket measuring device 1. FIG. FIG. 2 is a perspective view showing the appearance of the wand 11. FIG. FIG. 2 is a perspective view showing the appearance of the wand 11. FIG. It is a flowchart showing the operation of the periodontal pocket measuring device 1. It is a flowchart showing the operation of the periodontal pocket measuring device 1. It is a flowchart showing the operation of the periodontal pocket measuring device 1. It is a flowchart showing the operation of the periodontal pocket measuring device 1. It is a schematic diagram of teeth and gums. It is a schematic diagram of a tooth crown, a tooth root, and a gingiva.

Hereinafter, specific embodiments to which the present invention is applied will be described in detail with reference to the drawings.

FIG. 1 is a block diagram showing the configuration of a periodontal pocket measuring device 1 according to Embodiment 1 of the present invention. The periodontal pocket measuring device 1 includes a wand 11 and an information processing device 13.

The wand 11 is typically a cane-shaped device that can be grasped by medical staff such as dentists and dental hygienists (hereinafter simply referred to as dentists), and is Used to photograph shapes. FIG. 2 is a perspective view showing the appearance of the wand 11 in this embodiment. The wand 11 includes a scanner 111 and an air blower 113. In this example, a scanner 111 and an air blower 113 are built into the housing of the wand 11.

Scanner 111 includes a light source 1111, a sensor 1113, and a camera 1115. The light source 1111 emits light to a plurality of points on the surface of the object, and the reflected light is detected by the sensor 1113. Camera 1115 preferably captures a color image of the object surface. Note that the camera 1115 may be one that captures monochrome images.

Note that although FIG. 2 shows an example in which the light source 1111, sensor 1113, and camera 1115 are exposed near the tip of the housing of the wand 11 and are arranged in parallel, this arrangement may be changed as appropriate. For example, the light source 1111, the sensor 1113, or the camera 1115 may be arranged inside the housing of the wand 11 so as not to be visible from the outside. In this case, an optical path connecting the light source 1111, sensor 1113, or camera 1115 to the outside of the housing can be provided, and the design can be such that the emitted light or reflected light passes through this optical path.

The information processing device 13 includes a sensing data acquisition section 131, a model generation section 133, a recognition section 135, a measurement section 137, and a determination section 139.

The sensing data acquisition unit 131 acquires the detection result of the reflected light by the sensor 1113. Furthermore, an image of the surface of the object photographed by the camera 1115 is acquired. The image is preferably a color image, but may be a monochrome image.

The model generation unit 133 uses the detection results of the reflected light by the sensor 1113 to identify the distance from the scanner 111 to each point on the object surface by a known method such as the active wavefront sampling method or the confocal method. . Furthermore, based on this distance, the three-dimensional coordinates of each point on the object surface are specified. This set of points is called point cloud data.

Furthermore, the model generation unit 133 adds color information to the point cloud data using the color image of the object surface acquired by the camera 1115. Normally, point cloud data only has coordinate information, but by mapping 3D point cloud data onto a 2D camera image and associating the color information of the corresponding pixel with the point cloud data, point cloud data can be converted into point cloud data. Color information can be provided (see Non-Patent Document 1). This is called colored point cloud data.

Note that when the image acquired by the camera 1115 is monochrome, the model generation unit 133 adds monochrome color information (information indicating black and white shading) to the point cloud data. Hereinafter, point cloud data to which monochrome color information is added will also be referred to as colored point cloud data.

The model generation unit 133 outputs colored point group data or three-dimensional data (polygon data, CAD data, etc.) created based on the colored point group data as a model.

The sensing data acquisition unit 131 and the model generation unit 133 repeatedly perform sensing and model generation at regular intervals. By using a plurality of models generated in this way, various types of processing can be performed using temporal changes (such as movement) of the models.

The recognition unit 135 recognizes the crown, tooth root, and gingiva from the model output by the model generation unit 133. Typically, methods for distinguishing between teeth (including crowns and roots) and gums include a method that focuses on the movement of the tissue when the air blower 113 is injected, and a method that focuses on the color of the tissue. One method for distinguishing between a tooth crown and a tooth root is to focus on the color of the tissue. Regardless of whether motion or color is used, there are a discrimination method based on a threshold value and a discrimination method based on machine learning. The details will be described later.

The measurement unit 137 uses the recognition result of the recognition unit 135 to measure the depth of the periodontal pocket. The details will be described later.
The measurement unit 137 can output the measured periodontal pocket depth by displaying it on a screen, outputting it to a database or file, or printing it on a form.

The determining unit 139 determines the degree of progress of periodontal disease based on the information (for example, color, movement, etc.) possessed by the model output by the model generating unit 133 and the depth of the periodontal pocket measured by the measuring unit 137. . The details will be described later.
The determination unit 139 can output the determined degree of progress of periodontal disease by displaying it on a screen, outputting it to a database or file, or printing it on a form.

Note that the information processing device 13 includes hardware such as a processing device (CPU), a storage device, an input/output device, and a communication device. Each of the processing units (131, 133) described above is logically realized by the CPU executing a program stored in a storage device.

The air blower 113 injects compressed air from an injection port 1131. In the example of FIG. 2, an injection port 1131 of the air blower 113 is provided near the tip of the casing of the scanner 111. Although the direction of the injection port 1131 is preferably variable, it may be fixed. In any case, it is assumed that the direction of the injection port 1131 is set so that compressed air can be efficiently injected onto the target object. The air blower 113 injects compressed air to the target object when the scanner 111 photographs the target object (including before or during photographing). The injected compressed air scatters and removes saliva, blood, etc. present on the surface of the object. Thereby, the scanner 111 can acquire accurate reflected light from the surface of the object. Furthermore, when compressed air is injected into the pocket-like area (gingival sulcus) between the teeth and the gums, the pocket is expanded by the compressed air. Thereby, it is possible to easily image the teeth below the gingival margin without requiring measures such as exclusion.

The air blower 113 may be configured to automatically start jetting compressed air in conjunction with the operation of the scanner 111. Alternatively, the air blower 113 may include a button or the like for manually starting the injection of compressed air.

The air volume or air pressure of the air blower 113 may be fixed or variable. If it is variable, it is desirable that the information processing device 13 be able to acquire information indicating the air volume or air pressure at the time of injection. The information indicating the air volume or wind pressure may be a physical quantity expressed in units such as m3/min or Pa, or simply expressed in units of 1, 2, 3, . ．． It may be a set value indicating the magnitude of the air volume or wind pressure.

In the example of FIG. 2, the scanner 111 and the air blower 113 are integrated, but they may be configured to be separable. For example, as shown in FIG. 3, the wand 11 may be realized by externally attaching an air blower 113 to a conventional scanner 111 via an attachment 1132.

When the scanner 111 photographs the object, the air blower 113 sprays compressed air onto the object, thereby preventing the object from being wet with saliva, blood, etc., or being located below the gingival margin. An accurate model of the object can be obtained with one-handed operation, without requiring any measures such as suction or exclusion. Therefore, the time and effort of dentists and the like can be reduced compared to conventional methods. In addition, the physical burden on the patient can be reduced.

<Identification of tooth crown, tooth root, and gingiva>
(1) Method based on differences in vibration characteristics FIG. 5A is a schematic diagram of teeth and gums. When compressed air is injected into the pocket-like area between the teeth and gums, the compressed air forces the pocket open. Objects (teeth and gums) that are sprayed with compressed air vibrate, but the characteristics of the vibrations vary greatly depending on the area. That is, there should be a difference in amplitude and period between relatively soft gums and teeth (including crowns and roots) that are hard and fixed to bone.

Therefore, in this embodiment, the state around the teeth under the gum line is continuously photographed while injecting compressed air. In other words, a "video" is shot at a predetermined frame rate. Next, the captured video is analyzed to detect differences in the vibration characteristics of the object. Thereby, teeth and gums included in the model are distinguished. The recognition unit 135 performs processing to identify teeth and gums.

An example of the operation of the periodontal pocket measuring device 1 that discriminates between teeth (including crowns and roots) and gingiva based on differences in vibration characteristics will be described using the flowchart of FIG. 4A.

Step 1: Shooting a video The dentist or the like points the wand 11 at the subgingival margin and starts shooting. The air blower 113 starts operating, and the pockets between the teeth and gums are expanded by the jetted compressed air. The scanner 111 starts photographing and continuously photographs the intraoral shape of the teeth and subgingival region at predetermined intervals (frame rate). Thereby, the model generation unit 133 outputs a plurality of models that are continuous in time series. This is, so to speak, a three-dimensional "video." This “video” is stored in a storage device within the information processing device 13.

Step 2: Extraction of feature points The recognition unit 135 extracts feature points from each model included in the "video". Typically, all or part of the point group included in the model can be used as feature points.

Step 3: Calculating the movement of feature points The recognition unit 135 compares a plurality of temporally continuous models, finds corresponding feature points, and records the coordinates of the feature points in each frame. This process is performed for all frames. That is, the coordinate n1 of the feature point N in the model at time t1, the coordinate n2 of the feature point N in the model at time t2, the coordinate n3 of the feature point N in the model at time t3, etc. are sequentially specified. Then, a data set {n1, n2, n3...} indicating the movement of the feature point N is generated. Similar processing is performed for other feature points O, P, Q, . . . .

Step 4: Distinguish between teeth and gums based on characteristic point movement The recognition unit 135 detects a difference in the characteristic point movement. Typically, there are a discrimination method based on a threshold value and a discrimination method based on machine learning.

(Discrimination based on threshold)
The recognition unit 135 calculates an index for evaluating the movement of the feature point N based on the data set {n1, n2, n3...} indicating the movement of the feature point N. For example, if the feature point N is vibrating, it can be calculated using its amplitude and period as indicators. Indices are similarly calculated for other feature points O, P, Q, and so on.

The recognition unit 135 classifies feature points by applying a predefined threshold to the calculated index. For example, a label indicating gingiva is assigned to a feature point whose amplitude exceeds the threshold value X, and a label indicating teeth is assigned to a feature point whose amplitude is equal to or less than the threshold value X. Here, the threshold value may be a value determined in advance through a test or the like. For example, different threshold values may be defined depending on the patient's age and gender (assumed to be stored in advance in a storage area not shown), information indicating the air volume or air pressure of the air blower 113, and the like.

(Discrimination by machine learning)
The recognition unit 135 uses the dataset {n1, n2, n3...} indicating the movement of the feature points N as learning data and performs machine learning to classify the feature points into abutment teeth and gingiva. be able to. For example, the recognition unit 135 includes a machine learning unit 1351 that performs unsupervised learning. When the machine learning unit 1351 receives a data set indicating the movement of a large number of feature points as learning data, it automatically identifies differences in characteristics and forms a set (cluster) of feature points having the same characteristics. The machine learning unit 1351 assigns a label indicating gingiva to the feature points included in one cluster, and a label indicating an abutment tooth to the feature points included in the other cluster.

Note that the machine learning unit 1351 may classify the feature points into abutment teeth and gingiva using other known machine learning methods such as supervised learning and deep learning. For example, in supervised learning, in the learning phase, a large amount of known training data is used, which consists of a dataset showing the movement of a feature point and a label indicating whether the feature point is an abutment tooth or gingiva. The information is given to the machine learning unit 1351. Thereby, the machine learning unit 1351 gradually learns the correlation between the data set indicating the movement of the feature point and the label indicating whether the feature point is an abutment tooth or a gingiva. As learning progresses, the machine learning unit 1351 operates as an estimator that inputs an unknown data set indicating the movement of feature points and outputs a label with a high correlation to the feature points.

In addition, in the learning phase and the estimation phase, in addition to the data set indicating the movement of the feature points, for example, information indicating the age and gender of the patient, the air volume or air pressure of the air blower 113, etc. may be input. This makes it possible to generate a learning model that performs optimal estimation according to the patient's age and gender, the air volume or air pressure of the air blower 113, and the like.

(2) Method based on color difference FIG. 5B is a partially enlarged view of FIG. 5A. Teeth are divided into crowns and roots. Generally, the crown of the tooth exhibits a color close to white, which is a characteristic of enamel, and the root of the tooth exhibits a color close to cream, which is a characteristic of cementum. Also, the gums appear pink or red. In this embodiment, since colored point cloud data can be acquired, it is possible to identify the crown, root, and gingiva by focusing on the difference in color as described above. Note that when using a monochrome image, the color should be read as information indicating black and white shading.

An example of the operation of the periodontal pocket measuring device 1 that distinguishes the crown, root, and gingiva based on the difference in color will be described using the flowchart in FIG. 4B.

Step 1: Imaging A dentist or the like points the wand 11 at the subgingival margin and starts imaging. The air blower 113 starts operating, and the pockets between the teeth and gums are expanded by the jetted compressed air. The scanner 111 starts photographing, and the model generation unit 133 outputs a three-dimensional model showing the intraoral shape of the subgingival region. This model includes colored point cloud data. The model is stored in a storage device within the information processing device 13.

Step 2: Extraction of feature points The recognition unit 135 extracts feature points from the model. Here, the feature points include color information as well as coordinate information. Typically, all or part of the point group included in the model can be used as feature points.

Step 3: Distinguishing tooth crown, tooth root, and gingiva based on the color of the feature points The recognition unit 135 detects differences in the colors of the feature points. Typically, there are a discrimination method based on a threshold value and a discrimination method based on machine learning.

(Discrimination based on threshold)
The recognition unit 135 calculates an index for evaluating the color of the feature point N. For example, the RGB values of color information held by the feature point N can be used as an index.

The recognition unit 135 classifies feature points by applying a predefined threshold to the calculated index. For example, if each RGB value is r1<R≦r2, g1<G≦g2, and b1<B≦b2, it is a tooth crown, and r3<R≦r4, g3<G≦g4, and b3<B≦b4. If so, a label indicating the tooth root, and if r5<R≦r6, g5<G≦g6, and b5<B≦b6, a label indicating the gingiva is provided, respectively. Here, the threshold value may be a value determined in advance through a test or the like. For example, different threshold values may be defined depending on the age and gender of the patient.

Similar techniques include color clustering, color reduction or simplification. Color clustering can be realized by a known technique such as the K-means method. Although the colors of the feature points are actually different, by specifying the number of colors K and performing color clustering, the colors of the feature points can be classified into K colors. The recognition unit 135 assigns a label indicating the cluster (dental crown, tooth root, gingiva) to the feature points classified into each cluster.

(Discrimination by machine learning)
The recognition unit 135 can classify the feature points into crowns, roots, and gums by performing machine learning using the color information of the feature points N as learning data. For example, the recognition unit 135 includes a machine learning unit 1351 that performs unsupervised learning. When the machine learning unit 1351 receives a data set indicating the colors of a large number of feature points as learning data, it automatically identifies differences in the characteristics and forms a set (cluster) of feature points having the same characteristics. The machine learning unit 1351 assigns a label indicating a tooth crown to a feature point included in a color cluster close to white, a label indicating a tooth root to a feature point included in a color cluster close to cream color, and a label indicating a tooth root to a feature point included in a color cluster close to red. A label indicating gingiva is given to the included minutiae.

Note that the machine learning unit 1351 may use other known machine learning methods such as supervised learning and deep learning to classify feature points into crowns, roots, and gingiva. For example, in supervised learning, during the learning phase, a large amount of known training data is used, which is a set of data indicating the color of a feature point and a label indicating whether the feature point is a crown, root, or gingiva. The information is given to the machine learning unit 1351. Thereby, the machine learning unit 1351 gradually learns the correlation between the data indicating the color of the feature point and the label indicating whether the feature point is a crown, a tooth root, or a gingiva. As learning progresses, the machine learning unit 1351 operates as an estimator that inputs unknown data indicating the color of a feature point and outputs a label with a high correlation to the feature point.

Note that in the learning phase and the estimation phase, in addition to the data indicating the color of the feature points, information indicating the patient's age and gender, for example, may be input. This makes it possible to generate a learning model that performs optimal estimation according to the patient's age and gender.

Step 4: Determining tooth type The recognition unit 135 recognizes each tooth included in the feature point group and specifies the type of each tooth. That is, the type of each tooth (generally identified by a name such as central incisor, lateral incisor, canine, etc., or a number such as No. 1, No. 2, No. 3, etc.) is specified. This process can be realized by using a known technique such as that described in Patent Document 3, for example.

<Measurement of periodontal pocket depth>
The measurement unit 137 measures the depth of the periodontal pocket using the feature points classified by the recognition unit 135 into crowns, roots, and gingiva.
(1) Method for measuring the depth from the crown-tooth root boundary to the pocket bottom The measuring unit 137 measures the depth from the crown-tooth root boundary to the pocket bottom (see FIG. 5B), and measures this depth in the periodontal pocket. Output as depth. An example of the operation of the periodontal pocket measuring device 1 will be described using the flowchart in FIG. 4C.

Step 1: Identifying the tooth crown-tooth root boundary line The measuring unit 137 identifies the tooth crown-tooth root boundary line. Specifically, for example, a three-dimensional curved object is generated that indicates a boundary line between a group of feature points belonging to a tooth crown cluster and a group of feature points belonging to a tooth root cluster.

Step 2: Identifying the bottom of the pocket The measuring unit 137 identifies the bottom of the pocket. Specifically, for example, a three-dimensional curved object is generated that indicates a boundary line between a group of feature points belonging to a tooth root cluster and a group of feature points belonging to a gum cluster. Alternatively, the measurement unit 137 may simply generate a three-dimensional curved object indicating a line connecting the lowest points of the gums, and use this object as the pocket bottom.

Step 3: Setting measurement points Generally, periodontal pocket depth is measured at several measurement points per tooth. The measurement point here refers to the location where a probe was conventionally inserted into the gingival sulcus. Similarly, the measurement unit 137 sets a plurality of measurement points on the crown-tooth root boundary line. The number of measurement points to be set and the setting method are arbitrary.

Step 4: Measuring the depth of the periodontal pocket The measuring unit 137 measures the vertical distance (difference in height) between the measurement point set on the crown-tooth root boundary line and the point on the bottom of the pocket closest to the measurement point. is calculated and output as the periodontal pocket depth.

Note that the measurement point in step 3 may be provided on the bottom of the pocket. In this case, in step 4, the measurement unit 137 calculates the vertical distance (difference in height) between the measurement point set on the pocket bottom and the point on the tooth crown-tooth root boundary line closest to the measurement point, and Output as the depth of the circumferential pocket.
(2) Method of measuring the depth from the gingival crest to the pocket bottom The measuring unit 137 measures the depth from the gingival crest to the pocket bottom (see FIG. 5B) and outputs this as the depth of the periodontal pocket. . An example of the operation of the periodontal pocket measuring device 1 will be described using the flowchart in FIG. 4D.

Step 1: Identifying the Pocket Bottom The measuring unit 137 identifies the pocket bottom. Specifically, for example, a three-dimensional curved object is generated that indicates a boundary line between a group of feature points belonging to a tooth root cluster and a group of feature points belonging to a gum cluster. Alternatively, the measurement unit 137 may simply generate a three-dimensional curved object indicating a line connecting the lowest points of the gums, and use this object as the pocket bottom.

Step 2: Setting the gingival crest line The measuring unit 137 generates a three-dimensional curve object indicating a line (referred to as a gingival crest line) connecting the gingival crests (the highest points of the gums).

Step 3: Setting measurement points The measurement unit 137 sets a plurality of measurement points on the bottom of the pocket. The number of measurement points to be set and the setting method are arbitrary.

Step 4: Measurement of periodontal pocket depth The measurement unit 137 calculates the vertical distance (difference in height) between the measurement point set on the pocket bottom and the point on the gingival crest line closest to the measurement point. , output as periodontal pocket depth.

Note that the measurement point in step 3 may be provided on the gingival crest line. In this case, in step 4, the measuring unit 137 calculates the vertical distance (difference in height) between the measurement point set on the gingival crest line and the point on the bottom of the pocket closest to the measurement point, and Output as depth.

<Determination of progress of periodontal pockets>
The determination unit 139 determines the degree of progress of periodontal disease based on information possessed by the model (for example, the color of the gums, movement of the gums when air is injected, etc.) and the depth of periodontal pockets.

It is known that the depth of periodontal pockets tends to increase as periodontal disease progresses. In addition, when the gums become inflamed, they become congested and become reddish, and the gums tend to become loose and vibrate when air is injected. Conventionally, dentists have determined the degree of progress of periodontal disease by integrating these findings. However, if there are variations in the determination, it becomes difficult to accurately trace the results of treatment for, for example, periodontal disease. Therefore, in this embodiment, the degree of progression of periodontal disease is determined by machine learning to support treatment based on more objective data.

The machine learning unit 1351 can determine the degree of progress of periodontal disease using known machine learning techniques such as supervised learning and deep learning. For example, in supervised learning, in the learning phase, coordinates and color information of feature points, time series data indicating the movement of the feature points (see step 3 in Figure 4A), or indicators indicating the movement of the feature points (amplitude, period, frequency The machine learning unit 1351 generates a large number of known training data that is a set of data sets including coordinates of measurement points, depths of periodontal pockets at measurement points, etc.), and labels indicating the degree of progress of periodontal disease. give to Here, the label indicating the degree of progress of periodontal disease can be based on the diagnosis result by a dentist. For example, dentists consider a periodontal pocket depth of 3 mm or less to be considered "healthy," and a periodontal pocket depth of about 3 to 5 mm with red, swollen, and swollen gums to be considered "mild dental health." ``periodontal disease'', when the periodontal pocket depth is about 5-7 mm and the alveolar bone has begun to dissolve, ``moderate periodontal disease'', and when the periodontal pocket depth is 7 mm or more, bone resorption has progressed. If your teeth are loose, you may be diagnosed with ``severe periodontal disease.'' Such diagnostic results such as "healthy", "mild periodontal disease", "moderate periodontal disease", and "severe periodontal disease" can be used as labels in the learning phase.

Thereby, the machine learning unit 1351 gradually learns the correlation between various information included in the data set and the label indicating the degree of progress of periodontal disease. As the learning progresses, the machine learning unit 1351 operates as an estimator that inputs an unknown data set and outputs a label highly correlated with the data set, that is, the degree of progress of periodontal disease.

In addition to the information held by the model (for example, the color of the gingiva, the movement of the gingiva when air is injected, etc.) and the depth of the periodontal pocket, the determination unit 139 also collects information regarding the patient's age and tooth type. It may be used to determine the degree of progress of periodontal disease. It is known that the progress of gingival recession varies to some extent depending on the age of the patient and the type of teeth. By using this information, more appropriate judgments can be made according to the patient's age and tooth type.

In this case as well, the machine learning unit 1351 can determine the degree of progress of periodontal disease by using known machine learning methods such as supervised learning and deep learning, taking into account the age of the patient and the type of teeth. For example, in supervised learning, in the learning phase, the coordinates and color information of the feature points, time series data indicating the movement of the feature points or indicators indicating the movement of the feature points, the coordinates of the measurement point, the depth of the periodontal pocket at the measurement point , the type of tooth for which the measurement point was set (the one specified by the recognition unit 135 is used), the patient's age (assumed to be stored in advance in a storage area not shown), and periodontal disease. The machine learning unit 1351 is provided with a large number of known teacher data, each pairing with a label indicating the degree of progress. Here, the label indicating the degree of progress of periodontal disease can be based on the diagnosis result by a dentist.

Thereby, the machine learning unit 1351 gradually learns the correlation between the data set including information regarding the patient's age and tooth type, and the label indicating the degree of progress of periodontal disease. As the learning progresses, the machine learning unit 1351 operates as an estimator that inputs an unknown data set and outputs a label highly correlated with the data set, that is, the degree of progress of periodontal disease.

<Storage and use of measurement results and judgment results>
The measurement unit 137 can store the measured periodontal pocket depth measurement results in a storage area (not shown) together with, for example, the patient's identifier and the measurement date and time. Similarly, the determination unit 139 can save the determination result of the degree of progression of periodontal disease, together with the patient's identifier and the date and time of measurement, in a storage area (not shown). At this time, it may be saved together with camera images, sensing data, models, etc. used for measurement or determination.

If the above-mentioned measurements and judgments are performed every time a patient visits a dental clinic, historical information regarding the depth of each patient's periodontal pockets and the degree of progress of periodontal disease will be accumulated in chronological order. This history information can be used as evidence showing the patient's treatment progress. For example, the results of measuring the depth of periodontal pockets at each measurement point, the results of determining the degree of progress of periodontal disease, and the images taken by the camera from multiple past measurements are displayed side by side in a table, timeline, or the like. If it is numerical data such as the measurement results of periodontal pocket depth, its changes over time can be expressed using a graph. Based on this time-series information, a determination that the patient's periodontal disease is improving or worsening may be presented. These displays can be realized by outputting to a screen, file, form, etc., for example.

In the above-described embodiment, coordinates and color information of the feature points, time series data indicating the movement of the feature points (see step 3 in FIG. 4A), or indicators indicating the movement of the feature points (amplitude, period, frequency, etc.) are used. ), the coordinates of the measurement point, the depth of the periodontal pocket at the measurement point, etc. can be included in the dataset, but these are just examples, and only part of this information may be used. It is also possible to include information not shown in the data set.

According to the present embodiment, the periodontal pocket measuring device 1 can distinguish between teeth and gingiva according to the difference in their behavior by analyzing model data that is continuous over time. In addition, by using colored point cloud data, it is possible to distinguish between gums, tooth roots, and gingiva according to differences in color. Based on these analysis results, the depth of periodontal pockets can be accurately and easily estimated, and the degree of progression of periodontal disease can be objectively determined.

Note that the present invention is not limited to the above-described embodiments, and can be modified as appropriate without departing from the spirit of the present invention. For example, in the above-described embodiment, the periodontal pocket measuring device 1 acquires three-dimensional point group data, but it may also acquire, for example, two-dimensional image data. In this case, the model generation unit 133 can construct a three-dimensional model based on a plurality of two-dimensional image data. Note that since the method of constructing a three-dimensional model from two-dimensional images is a known technique, a detailed explanation will be omitted.

Further, when acquiring point cloud data, the periodontal pocket measuring device 1 may additionally acquire information regarding temperature and the like in addition to color information (including information regarding black and white shading). In this case, for example, if the machine learning unit 1351 uses this additional information as learning data, it is possible to improve the discrimination accuracy.

1 Periodontal pocket measurement device 11 Wand 111 Scanner 1111 Light source 1113 Sensor 1115 Camera 113 Air blower 1131 Nozzle 1132 Attachment 13 Information processing device 131 Sensing data acquisition section 133 Model generation section 135 Recognition section 1351 Machine learning section 137 Measurement section 139 Judgment section

Claims (8)

A scanner that photographs the inside of the oral cavity and obtains the shape of the oral cavity,
an air blower that injects compressed air to the object to be imaged when the scanner performs the imaging;
an information processing device;
The scanner acquires point cloud data indicating the intraoral shape while the air blower injects the compressed air to the imaging target;
The information processing device includes:
a sensing data acquisition unit that receives the point cloud data from the scanner;
a model generation unit that generates a model based on the point cloud data;
a recognition unit that identifies gums, tooth roots, and gingiva based on the characteristics of the model;
a measurement unit that measures the depth of periodontal pockets based on the identification results of the gums, tooth roots, and gingiva;
has,
Periodontal pocket measuring device. The periodontal pocket measuring device according to claim 1, further comprising a determination unit that determines the degree of progress of periodontal disease based on at least the depth of the periodontal pocket. The scanner continuously acquires the plurality of point cloud data,
The periodontal pocket measuring device according to claim 1, wherein the recognition unit identifies teeth and gums based on characteristics of changes in position of the point cloud data over time. the scanner acquires the point cloud data having color information;
The periodontal pocket measuring device according to claim 1, wherein the recognition unit identifies the gums, tooth roots, and gingiva based on color information of positions of the point cloud data. a sensing data acquisition step of receiving point cloud data indicating the intraoral shape obtained by imaging the intraoral cavity with the scanner while the air blower injects compressed air to the imaging target;
a model generation step of generating a model based on the point cloud data;
a recognition step of identifying gums, roots, and gingiva based on features of the model;
A periodontal pocket measuring method, comprising: a measuring step of measuring the depth of a periodontal pocket based on the identification results of the gums, tooth roots, and gingiva. The periodontal pocket measuring method according to claim 5, further comprising a determination step of determining the degree of progress of periodontal disease based on at least the depth of the periodontal pocket. A program for causing a computer to execute the method according to claim 5. A program for causing a computer to execute the method according to claim 6.

Images