KR102322634B1 - Method and apparatus for matching of images using tooth object - Google Patents

Abstract

The present invention relates to an image registration method and apparatus using a tooth object. An image registration method using a tooth object according to an embodiment of the present invention comprises: generating a plurality of reference points spaced apart from each other in an oral scan image and a computed tomography (CT) image of a registration target; and using the reference point (first reference point) of the oral scan image and the reference point (second reference point) of the CT image to match the oral scan image and the CT image of the target to be matched; including, the first and second reference points includes a reference point for one frontmost tooth of the anterior region and a reference point for two teeth on both rearmost sides of the posterior region, and the first reference point is derived from an object that is a simplified shape of a tooth.

Description

Image matching method and apparatus using a tooth object {Method and apparatus for matching of images using tooth object}

The present invention can perform image registration between an oral scan image and a CT image using an object of each tooth in an oral scan image and a computed tomography (CT) image of the inside of the oral cavity. It relates to a method and apparatus.

In the field of computer vision, when the same object is photographed at different times, measurement methods, or viewpoints, images having different coordinate systems are obtained. Image registration refers to processing for displaying these different images in one coordinate system.

In particular, in the field of dentistry, image registration between an oral scan image and a CT image is performed before a procedure such as an implant. At this time, the matched image can be used as important data for determining the optimal implant position by allowing the position of the bone tissue and the neural tube to be identified.

However, in a conventional method for image registration between an oral scan image and a CT image, image registration is performed using a marker of each image manually designated by a user, or the distance between all vertices included in each image is compared. and image registration was performed. As a result, the conventional method has a problem in that not only the speed of image registration is reduced due to a large load, but also the image registration is inaccurate due to inaccurate characteristics of manual markers and vertices.

US 10-1913586 B

In order to solve the problems of the prior art as described above, the present invention provides a method for quickly and accurately performing image registration between an oral scan image and a CT image by using the object of each tooth in an oral scan image and a CT image of the inside of the oral cavity. It is an object to provide a method and apparatus.

However, the problems to be solved by the present invention are not limited to the problems mentioned above, and other problems not mentioned will be clearly understood by those skilled in the art from the following description.

In an image registration method using a tooth object according to an embodiment of the present invention for solving the above problems, (1) a plurality of spaced apart from each other in an oral scan image and a computed tomography (CT) image of a registration target generating step of generating each reference point of , (2) using the reference point of the oral scan image (first reference point) and the reference point of the CT image (second reference point) to register the oral scan image and the CT image of the registration target; include

The first and second reference points may include a reference point for one frontmost tooth of the anterior region and a reference point for two teeth on both rearmost sides of the posterior region.

The first reference point may be derived from an object that is a simplified shape of a tooth.

The generating step includes (1) generating a learning model by learning with learning data that forms an object for each tooth in multiple directions of the oral scan image of the learning target, (2) matching target using the generated learning model The method may include detecting an object for a plurality of directions of each tooth in the oral scan image of , (3) selecting a reference object from among the detected objects, and generating a first reference point from the selected reference object.

In the generating step, (1) the model generating step includes n teeth (target teeth) spaced apart from each other, which are some teeth in multiple directions of the oral scan image of the learning target (where n is a natural number equal to or greater than 3). Generating a learning model trained with the formed learning data, (2) Object detection step of detecting objects in multiple directions of each target tooth in the oral scan image of the matching target using the generated learning model, (3) The method may include selecting the detected object as a reference object, and generating a first reference point from the selected reference object.

The reference object may include an object for one frontmost tooth of the anterior region and an object for two teeth on both rearmost sides of the posterior region.

The generating may further include selecting a center point of the selected reference object as the first reference point.

An image registration apparatus according to an embodiment of the present invention includes (1) a storage unit storing an oral scan image and a computed tomography (CT) image of a matching target, and (2) storing the oral scan image and CT image After generating a plurality of reference points spaced apart from each other, the control unit for registering the oral scan image and the CT image of the registration target by using the reference point (first reference point) of the oral scan image and the reference point (second reference point) of the CT image do.

The first and second reference points may include a reference point for one frontmost tooth of the anterior region and a reference point for two teeth on both rearmost sides of the posterior region.

The first reference point may be derived from an object that is a simplified shape of a tooth.

The control unit uses a learning model generated by learning with learning data that forms an object for each tooth in multiple directions of the oral scan image and the CT image of the learning target, and in the oral scan image of the matching target, multiple directions of each tooth After detecting an object for , a first reference point may be generated from the selected reference object by selecting a reference object from among the detected objects.

The control unit learns with learning data that forms each object for n teeth (target teeth) spaced apart from each other, which are some teeth in multiple directions of the oral scan image and CT image of the learning target (where n is a natural number equal to or greater than 3) By using the learning model created by can

The present invention configured as described above performs image registration between the oral scan image and the CT image using an object of each tooth that can be quickly and accurately extracted from the oral scan image and the CT image of the oral cavity, thereby speeding up the image registration. and improved accuracy.

1 is a block diagram showing an image matching apparatus 100 according to an embodiment of the present invention.
2 is a flowchart illustrating an image registration method using a tooth object according to an embodiment of the present invention.
3 shows a dental area comprising an anterior region FA and a posterior region BA.
4 is a more detailed flowchart of S100 of an image registration method using a tooth object according to an embodiment of the present invention.
5 is a diagram illustrating extraction from a region of interest (ROI) from the first learning target oral scan image.
6 shows a state in which objects for four directions are set in a region of interest (ROI) of the oral scan image as a second learning target.
7 illustrates an object detected in an oral scan image of first to fourth matching targets.
8 is a view illustrating a 3D object detected in various directions in an oral scan image of a fifth matching target.
9 and 10 show an image registration process between an oral scan image and a CT image of a registration target.
11 shows a state after image registration between an oral scan image and a CT image of a registration target.

The above object and means of the present invention and its effects will become more apparent through the following detailed description in relation to the accompanying drawings, and accordingly, those of ordinary skill in the art to which the present invention pertains can easily understand the technical idea of the present invention. will be able to carry out In addition, in the description of the present invention, if it is determined that a detailed description of a known technology related to the present invention may unnecessarily obscure the gist of the present invention, the detailed description thereof will be omitted.

The terminology used herein is for the purpose of describing the embodiments, and is not intended to limit the present invention. In this specification, the singular form also includes the plural as the case may be, unless otherwise specified in the text. In this specification, terms such as "include", "provide", "provide" or "have" do not exclude the presence or addition of one or more other elements other than the mentioned elements.

In this specification, terms such as “or” and “at least one” may indicate one of the words listed together, or a combination of two or more. For example, “or B” and “at least one of B” may include only one of A or B, and may include both A and B.

In the present specification, descriptions according to “for example” and the like may not exactly match the information presented, such as recited properties, variables, or values, tolerances, measurement errors, limits of measurement accuracy, and other commonly known factors. The embodiments of the present invention according to various embodiments of the present invention should not be limited by effects such as modifications including .

In this specification, when it is described that a certain element is 'connected' or 'connected' to another element, it may be directly connected or connected to the other element, but other elements may exist in between. It should be understood that there may be On the other hand, when it is mentioned that a certain element is 'directly connected' or 'directly connected' to another element, it should be understood that there is no other element in the middle.

In this specification, when it is described that a certain element is 'on' or 'adjacent' to another element, it may be directly in contact with or connected to the other element, but another element may exist in the middle. It should be understood that On the other hand, when it is described that a certain element is 'on top' or 'directly' of another element, it may be understood that another element does not exist in the middle. Other expressions describing the relationship between elements, for example, 'between' and 'directly between', etc. can be interpreted similarly.

In this specification, terms such as 'first' and 'second' may be used to describe various components, but the components should not be limited by the above terms. In addition, the above terms should not be construed as limiting the order of each component, and may be used for the purpose of distinguishing one component from another. For example, a 'first component' may be termed a 'second component', and similarly, a 'second component' may also be termed a 'first component'.

Unless otherwise defined, all terms used herein may be used with meanings commonly understood by those of ordinary skill in the art to which the present invention pertains. In addition, terms defined in a commonly used dictionary are not to be interpreted ideally or excessively unless specifically defined explicitly.

Hereinafter, a preferred embodiment according to the present invention will be described in detail with reference to the accompanying drawings.

1 is a block diagram showing an image matching apparatus 100 according to an embodiment of the present invention.

The image registration apparatus 100 according to an embodiment of the present invention is an electronic device that performs image registration between an oral scan image and a CT image of the inside of the oral cavity.

The oral scan image is an image that provides information on the shape of the crown portion of the tooth exposed to the outside and the shape of the gum around the tooth. At this time, the oral scan image is obtained by directly scanning the inside of the recipient's oral cavity through an oral scanner, etc. It is also possible, and the scan image of the impression model can be inverted and used as an oral scan image.

A CT image is an image taken through a computed tomography (CT) device using radiation. That is, the CT image may indicate the distribution of internal tissues of the oral cavity, such as the crown, the root, and the alveolar bone, and information on the bone density, etc., based on the transmittance of the radiation.

Referring to FIG. 1 , an image matching apparatus 100 according to an embodiment may include a communication unit 110 , an input unit 120 , a display unit 130 , a storage unit 140 , and a control unit 150 .

The communication unit 110 is configured to communicate with an external device such as an image acquisition device (not shown) and a server (not shown), and may receive image data. For example, the communication unit 110 is 5 ^th generation communication (5G), long term evolution-advanced (LTE-A), long term evolution (LTE), Bluetooth, bluetooth low energe (BLE), near field communication (NFC). It is possible to perform wireless communication such as cable communication and the like, and can perform wired communication such as cable communication.

In this case, the image data may include oral scan image data, CT image data, and the like.

The input unit 120 generates input data in response to a user's input. The input unit 120 includes at least one input means. For example, the input unit 120 includes a keyboard, a keypad, a dome switch, a touch panel, a touch key, a mouse, and a menu button ( menu button) and the like.

The display unit 130 displays display data according to the operation of the preprocessor 100 . Such display data may include image data. For example, the display unit 130 may include a liquid crystal display (LCD), a light emitting diode (LED) display, an organic light emitting diode (OLED) display, and a micro electromechanical system (MEMS). mechanical systems) displays and electronic paper displays. Also, the display unit 130 may be combined with the input unit 120 to be implemented as a touch screen or the like.

The storage unit 140 stores various types of information and programs necessary for the operation of the image matching apparatus 100 . For example, the storage unit 140 may store image data received from an image acquisition device, etc., and an algorithm related to an image matching method using a tooth object according to an embodiment of the present invention, which will be described later. Also, the storage unit 140 may store the learning model.

The controller 150 performs image matching between the CT image and the oral scan image received from the image acquisition device or server or pre-stored in the storage unit 140 . To this end, the controller 150 may receive image data from an image acquisition device or a server and store the received image data in the storage unit 140 . Also, the controller 150 may control the operations of the communication unit 110 , the input unit 120 , the display unit 130 , and the storage unit 140 .

Hereinafter, an image registration method using a tooth object according to an embodiment of the present invention controlled by the controller 150 will be described.

2 is a flowchart illustrating an image registration method using a tooth object according to an embodiment of the present invention.

Referring to FIG. 2 , the image registration method using a tooth object according to an embodiment of the present invention may include S100 and S200 for performing image processing on image data.

First, in S100 , the controller 150 generates a plurality of reference points spaced apart from each other in the oral scan image and the CT image of the matching target, respectively. That is, the controller 150 generates a reference point (hereinafter, referred to as a “first reference point”) in the oral scan image to be matched, and a reference point (hereinafter, referred to as a “second reference point”) in the CT image to be matched.

Thereafter, in S200 , the controller 150 performs image registration by changing the matching target oral scan image or the matching target CT image using the first reference point and the second reference point so that the reference points coincide with each other.

In this case, the reference point is a point indicating a position with respect to a specific tooth (eg, a position of a center point of the specific tooth), and is used during image registration. That is, in S200 , the control unit 150 changes the angle, size, location, etc. of the target oral scan image and the CT image to be matched so that the specific first reference point and the specific second reference point corresponding thereto coincide with each other, so that the Image registration can be performed.

3 shows a dental area comprising an anterior region FA and a posterior region BA.

Referring to FIG. 3 , an oral scan image and a CT image of the registration object include a tooth region indicating the shape of the teeth. In this case, the tooth area includes an anterior region FA located in the front and a posterior region BA located in the rear of the anterior region FA. For example, the anterior area FA may be an area in which teeth 1 to 3 are located, and the posterior area BA may be an area in which teeth 4 to 8 are located.

At this time, the reference point is a reference point (hereinafter, referred to as “forward reference point”) for one frontmost tooth FT of the anterior region FA, and two teeth (BT1, BT2) on both rearmost sides of the posterior region BA. Reference points (hereinafter, referred to as “rearmost reference points”) may be included, respectively. That is, among the rearmost reference points, one is a reference point for one tooth selected among the teeth on the right side of the posterior area BA, and the other is a reference point for the other selected tooth among the teeth on the left side of the posterior area BA. is the reference point.

These forward and rearmost reference points are generated in each of the registration object oral scan image and the registration object CT image. As a result, in each image, one frontmost reference point and two rearmost reference points form a triangular vertex, and the angle, size, position, etc. of the images are changed during image registration between the oral scan image to be matched and the CT image to be matched. It can provide a simpler and more accurate standard for

In this case, the second reference point may be easily derived manually or through various algorithms from the 3D coordinate information of the CT image to be matched, which may indicate information such as the structure, size, and position of the inside of the tooth. In this case, the derived second reference point may be a point indicating the position of the center point of the one frontmost tooth FT and the two rearmost teeth BT1 and BT2. Of course, the second reference point may be derived using a method of deriving (detecting) the first reference point, which will be described later. In this case, the description of the 'oral scan image' may be replaced with the description of the 'CT image'.

Meanwhile, it is necessary to derive the first reference point corresponding to the derived second reference point. However, since the first reference point must be derived from an oral scan image indicating information on the surface of the tooth, the accuracy of the manual or conventional algorithm (using curvature information in the oral scan image) is inevitably reduced.

Accordingly, the present invention derives the first reference point using the object OB, which is a simplified shape of a tooth. This will be described in more detail according to S101 to S104 to be described later.

4 is a more detailed flowchart of S100 of the image registration method using a tooth object according to an embodiment of the present invention, and FIG. 5 shows a state of extracting the ROI from the first learning target oral scan image.

Referring to FIG. 5 , in S101 , the controller 150 extracts a region of interest (ROI) including teeth from an oral scan image (hereinafter, referred to as a “learning oral scan image”) of a learning target. That is, the learning oral scan image may include a tooth region and an extra-tooth region, and in S101 , the controller 150 may extract the tooth region as a region of interest (ROI).

Thereafter, in S102 , the controller 150 generates a learning model by learning from training data that forms the object OB1 for each tooth in a plurality of directions of the extracted region of interest ROI. In this case, the controller 150 may learn the prepared learning data using a machine learning technique.

For example, machine learning techniques are supervised learning techniques, artificial neural networks, Boosting, Bayesian statistics, Decision tree, Gaussian process regression, Nearest neighbor algorithm, Support vector machine, Random forests, Symbolic machine learning, Ensembles of It may be one of techniques such as classifiers and deep learning.

That is, the learning data is an input value of the extracted learning oral scan image of the region of interest (ROI), and an object (OB) formed for each tooth in the image is a result value ( target value). At this time, the object OB1 is set in various directions in the region of interest (ROI) of the learning oral scan image, and has various shapes (the shape of the corresponding tooth in the corresponding direction) in which the shape of the corresponding tooth is simplified in the corresponding direction to correspond to the shape of each tooth. It may be a shape that covers the area but is simpler than the shape of the tooth), that is, a shape such as a circle or a polygon. In addition, the result value of the learning data for the object OB1 may include position information, center point information, and size information occupied by the object OB1 in the region of interest ROI.

In addition, the learning model is a model supervised by a machine learning technique using training data, and includes a rule function for matching the input value and the result value.

6 shows a state in which an object OB1 for four directions is set in a region of interest (ROI) of an oral scan image to be learned. That is, FIG. 6(a) shows a planar direction, FIG. 6(b) shows a front direction, FIG. 6(c) shows a left direction, and FIG. 6(d) shows a right direction, respectively.

For example, as shown in FIG. 6 , the object OB1 for four directions, ie, plane, front, left, and right directions, may be set in the region of interest (ROI) of the training oral scan image. In this case, the setting of the object OB1 may be set by the user through the input unit 120 . As a result, the learning data may include the object OB1 in at least two directions (specific directions) for each tooth. In this case, the specific direction may include a planar direction and an out-of-plane direction (a front direction, a left direction, or a right direction).

That is, the object OB1 may be formed in the plane direction and the front direction of the region of interest ROI for teeth 1 to 3 seen in the front direction of the region of interest ROI. In addition, the object OB1 may be formed in the left direction and the front direction of the ROI for the left 4th to left 8th teeth seen in the left direction of the ROI. In addition, the object OB1 may be formed in the right direction and the front direction of the ROI with respect to the right fourth to right eighth teeth seen in the right direction of the ROI. Also, the object OB1 for all teeth may be formed in the plane direction of the region of interest ROI as described above.

The object OB1 in the planar direction is a shape obtained by simplifying the shape of the corresponding tooth in the planar direction, and the object OB1 in the out-of-plane direction is a shape obtained by simplifying the shape in the out-of-planar direction of the corresponding tooth. Accordingly, the object OB1 in the plane direction may act as a medium providing information about the plane of the corresponding tooth, and the object OB1 in the out-of-plane direction provides information on one side (height, etc.) of the corresponding tooth. It can serve as a medium to provide.

For example, the object OB1 in the plane direction with respect to the teeth of the mandible may act as a medium providing information on the upper surface of the corresponding tooth, and the object OB1 in the out-of-plane direction with respect to the teeth of the mandible corresponds to the corresponding object OB1. It may act as a medium for providing information on one side of the tooth. Similarly, the object OB1 in the plane direction with respect to the maxillary tooth may act as a medium for providing information on the lower surface of the corresponding tooth, and the object OB1 in the out-of-plane direction with respect to the maxillary tooth is the corresponding tooth. It can act as a medium that provides information about one aspect.

Thereafter, in S103 , the controller 150 detects the object OB2 for multiple directions of each tooth in the oral scan image of the matching target using the learning model generated in S102 . That is, the controller 150 may input the matching target oral scan image as an input value to the learning model, and as a result, the learning model may output the object OB2 for the matching target oral scan image as the result value. have.

7 shows a state in which the object OB2 is detected in the oral scan image of the first to fourth detection targets, and FIG. 8 is a view in which the 3D object OB3 is detected in various directions in the oral scan image of the fifth detection target. shows the appearance of

Referring to FIG. 7 , the learning model may output (detect) an object OB2 corresponding to a plurality of directions in S102 from an oral scan image to be matched.

That is, for teeth 1 to 3 seen in the front direction of the oral scan image to be matched, the corresponding object OB2 may be detected in the planar direction and the front direction of the oral scan image to be matched, respectively. In addition, the object OB2 may be detected in the left direction and the front direction of the registration target oral scan image with respect to left 4th to left 8th teeth seen in the left direction of the oral scan image to be matched. In addition, the object OB1 may be detected in the right direction and the front direction of the registration target oral scan image with respect to right 4th to right 8th teeth seen in the right direction of the oral scan image to be matched. Also, the object OB2 for all teeth may be detected in the plane direction of the region of interest ROI as described above.

Meanwhile, in S103 , the controller 150 may generate (detect) the three-dimensional object OB3 by using two or more objects OB2 for each detected tooth. That is, the controller 150 may generate a 3D shape including two or more objects OB2 for each detected tooth as a surface, and detect the 3D shape as the 3D object OB3 .

At this time, the object OB2 generated by the learning model has various shapes that simplify the shape of the corresponding tooth in the corresponding direction to correspond to the shape of each tooth (a shape that covers the area of the corresponding tooth in the corresponding direction but is simpler than the shape of the tooth), that is, It may have a shape such as a circle or a polygon. In addition, the three-dimensional object OB3 has various three-dimensional shapes in which the three-dimensional shape of the corresponding tooth is simplified to correspond to the three-dimensional shape of each tooth (a shape that covers the volume of the corresponding tooth but is simpler than the three-dimensional shape of the tooth), that is, a cylinder. , may be in the form of an elliptical column, a polygonal column, a truncated cone, or a polygonal truncated pyramid.

For example, as shown in FIG. 8 , when the three-dimensional object OB3 is a rectangular parallelepiped, the controller 150 has two objects OB2 for a certain tooth as a first surface and a second surface. By generating a rectangular parallelepiped having virtual surfaces perpendicular to the first surface and the second surface as the remaining surfaces, the cubical-shaped 3D object OB3 for the corresponding tooth may be detected.

That is, the object OB2 in the planar direction represents the shape of the corresponding tooth in the planar direction, and the object OB2 in the out-of-plane direction represents the shape in the out-of-planar direction of the corresponding tooth. Accordingly, the object OB2 in the plane direction may act as a medium providing information on the plane of the corresponding tooth, and the object OB2 in the out-of-plane direction provides information on one side (height, etc.) of the corresponding tooth. It can serve as a medium to provide.

As a result, the control unit 150 has the object OB2 in the plane direction as the upper surface or the lower surface of the corresponding tooth and the object OB2 in the out-of-plane direction as the first side of the tooth. can be detected. In this case, the controller 150 may add surfaces other than a surface formed by the object OB2 in the planar direction and the object OB2 in the out-of-plane direction as virtual surfaces. That is, the controller 150 may add one or more virtual surfaces perpendicular to the object OB2 in the planar direction as the other side, and convert the virtual surface parallel to the object OB2 in the planar direction to the other plane. (If the corresponding tooth is included in the mandible, the lower side, if the corresponding tooth is included in the upper jaw, the single side) can be added.

Thereafter, in S103, the controller 150 may extract location information (location coordinates in the oral scan image), center point information (center point coordinates in the oral scan image) and size information of each detected object OB2, OB3. have. In this case, the location information, center point information, and size information of the object OB2 may be output together with the object OB2 as a result value for the input value of the learning model.

However, the controller 150 uses position information, center point information, and size information for two or more related objects OB2 used to generate each three-dimensional object OB3, Location information, center point information, and size information can be extracted.

For example, as shown in FIG. 8 , when the three-dimensional object OB3 is a rectangular parallelepiped, the control unit 150 controls position information and center point information of the objects OB2 constituting the first and second surfaces of the rectangular parallelepiped. and using the size information, position information, center point information, and size information of the corresponding 3D object OB3 may be extracted.

Thereafter, the controller 150 may display the detected objects OB2 and OB3 on the matching target oral scan image. That is, the controller 150 may display the respective objects OB2 and OB3 on the matching target oral scan image by using the position information, center point information, and size information of each detected object OB2 and OB3 . However, in order to make the division of each tooth more clear, as shown in FIG. 6 , the controller 150 may display the detected objects OB2 and OB3 in different colors for each tooth on the matching target oral scan image. have.

In this case, the controller 150 displays at least one of location information, center point information, and size information for each detected object OB2 and OB3 and each detected object OB2 and OB3 together on the matching target oral scan image. can do.

Meanwhile, although FIGS. 5 to 8 show only the teeth of the mandible, the present invention is not limited thereto, and the object detection operation according to the present invention may be equally applied to the teeth of the upper jaw.

9 and 10 show an image registration process between an oral scan image and a CT image of a registration target. At this time, in FIGS. 9 and 10 , the dotted rectangular parallelepiped indicates the regions of the frontmost tooth FT and the rearmost tooth BT1 and BT2 in each image. That is, FIGS. 9 and 10 show a process of image registration so that the first reference point and the second reference point in each image are matched with each other. Also, FIG. 11 shows a state after image registration between an oral scan image and a CT image of a matching target.

Thereafter, in S104 , the controller 150 selects a reference object from among the detected objects OB2 and OB3 and generates a first reference point from the selected reference object. At this time, the reference object is an object for one frontmost tooth FT of the anterior region FA (hereinafter, referred to as “anteriormost object”), and two teeth (BT1, BT2) on both rearmost sides of the posterior area BA. ) for (hereinafter, referred to as “the rearmost object”) may be included. Also, the controller 150 may select a center point (a center point of a surface or a volume) of the reference object as the first reference point.

The present invention can easily detect the objects OB2 and OB3 corresponding to each tooth in the oral scan image of the inside of the oral cavity, thereby improving the detection time and efficiency.

In addition, the present invention can provide extraction information such as location, center point, and size information of each detected object (OB2, OB3), and as a result, the extracted information can be utilized as a reference point for image registration with a CT image. In addition, the accuracy of the image registration can be improved because the accuracy of the extracted information is high.

That is, referring to FIGS. 9 and 10 , during image registration with a CT image, center point information on a plurality of objects (OB2, OB3) that can be quickly extracted can be used as the first reference point of image registration, The information on the first reference point is more accurate than the information according to the conventional tooth detection method using curvature information, and as a result, as shown in FIG. 11 , the speed and accuracy of image registration can be improved accordingly.

Meanwhile, in S100 , the controller 150 may control the learning model to output the frontmost object and the rearmost object as output values thereof.

In this case, in S102, the controller 150 controls n teeth (hereinafter, referred to as “target teeth”) spaced apart from each other, which are some teeth in multiple directions of the extracted region of interest (ROI) (where n is a natural number equal to or greater than 3) A learning model is created by learning according to the machine learning technique with the learning data that forms each object (OB1) for . That is, the target tooth includes an anteriormost tooth FT and a rearmost tooth BT. Accordingly, the learning model is learned using the learning data that forms the object OB1 corresponding to the frontmost object and the rearmost object. In addition, in S103, the control unit 150 uses the learning model generated in S102 to determine An object OB2 for multiple directions of each target tooth is detected from the oral scan image. Furthermore, the three-dimensional object OB3 is detected using this object OB2. As a result, in S104 , the controller 150 selects the detected objects OB2 and OB3 as reference objects, and generates a first reference point from the selected reference object. Other than that, it is the same as the contents of S101 to S104 described above.

In the detailed description of the present invention, although specific embodiments have been described, various modifications are possible without departing from the scope of the present invention. Therefore, the scope of the present invention is not limited to the described embodiments, and should be defined by the following claims and their equivalents.

100: preprocessor 110: communication unit
120: input unit 130: display unit
140: storage unit 150: control unit
BA: posterior area BT: rearmost tooth
FA: Anterior region FT: Anteriormost tooth
OB: object ROI: region of interest

Claims (8)

An image registration method performed by an electronic device, comprising:
generating a plurality of reference points spaced apart from each other in an oral scan image and a computed tomography (CT) image of a matching target; and
using the first reference point, which is a reference point of the oral scan image, and the second reference point, which is the reference point of the CT image, to register the oral scan image and the CT image to be matched;
The first and second reference points are the reference point for one tooth in the anterior region, which is the region where the teeth 1 to 3 are located, and the two teeth on both sides in the posterior region, which is the region where the teeth 4 to 8 are located. including reference points for each,
The first reference point is derived from an object having a simplified shape for a tooth,
The generating step is
Objects in two or more different directions for each tooth, the plane direction and the out-of-plane direction, are learned using the learning data including the object for each tooth in multiple directions of the oral scan image of the learning target. Learning by using the learning data including the step of generating a learning model;
detecting objects in multiple directions of each tooth from a target oral scan image that is an oral scan image of a matching target by using the generated learning model;
generating a three-dimensional object including each object in two or more different directions for each detected tooth as a surface; and
selecting a reference object from among the generated 3D objects to generate a first reference point from the selected reference object;
includes,
The learning data includes an oral scan image of a learning target as an input value, and an object formed for each tooth in the image as a result value forming a set with respect to the input value,
The result value of the learning data includes a simplified object in the plane and front directions of the region of interest in the case of teeth 1 to 3, respectively, and in the case of teeth 4 to 8 on the left, in the left and plane directions of the region of interest, respectively. Includes a simplified object, and in the case of right 4th to right 8th teeth, includes a simplified object in the right and planar directions of the region of interest, respectively,
In the detecting step, by inputting the target oral scan image to the learning model as an input value, an object for multiple directions of each tooth, which is a result value for the target oral scan image, is detected, but in the case of teeth 1 to 3, the target Objects are detected in the plane and front directions of the oral scan image, and in the case of left 4th to left 8th teeth, the object is detected in the left and planar directions of the target oral scan image, respectively, and the right 4th to right 8th teeth An image registration method that detects objects in the right and planar directions of the target oral scan image, respectively. delete delete According to claim 1,
The reference object includes a three-dimensional object for one frontmost tooth of the anterior region and a three-dimensional object for two teeth on both rearmost sides of the posterior region. According to claim 1,
The generating step may further include selecting a center point of the selected reference object as the first reference point. delete delete a storage unit storing an oral scan image and a computed tomography (CT) image of the matching target; and
After generating a plurality of reference points spaced apart from each other in the stored oral scan image and CT image, the oral scan image and CT image of the matching target are used using the first reference point, which is the reference point of the oral scan image, and the second reference point, which is the reference point of the CT image. Including; a control unit for matching
The first and second reference points are the reference point for one tooth in the anterior region, which is the region where the teeth 1 to 3 are located, and the two teeth on both sides in the posterior region, which is the region where the teeth 4 to 8 are located. including reference points for each,
The first reference point is derived from an object having a simplified shape for a tooth,
The control unit is
It is learned by a machine learning technique using learning data including objects for each tooth in multiple directions of the oral scan image and CT image of the learning target, but in two or more different directions for each tooth, the planar direction and the out-of-plane direction. Using a learning model generated by learning using learning data including the object of
Thereafter, a three-dimensional object including each object in two or more different directions for each detected tooth as a surface is generated,
Thereafter, by selecting a reference object from among the generated 3D objects, a first reference point is generated from the selected reference object,
The learning data includes an oral scan image of a learning target as an input value, and an object formed for each tooth in the image as a result value forming a set with respect to the input value,
The result value of the learning data includes a simplified object in the plane and front directions of the region of interest in the case of teeth 1 to 3, respectively, and in the case of teeth 4 to 8 on the left, in the left and plane directions of the region of interest, respectively. Includes a simplified object, and in the case of right 4th to right 8th teeth, includes a simplified object in the right and planar directions of the region of interest, respectively,
In the detecting step, by inputting the target oral scan image to the learning model as an input value, an object for multiple directions of each tooth, which is a result value for the target oral scan image, is detected, but in the case of teeth 1 to 3, the target Objects are detected in the plane and front directions of the oral scan image, and in the case of left 4th to left 8th teeth, the object is detected in the left and planar directions of the target oral scan image, respectively, and the right 4th to right 8th teeth An image matching device that detects objects in the right and planar directions of the target oral scan image, respectively.

Images