WO2022108082A1 - System and method for segmenting teeth in ct image - Google Patents

System and method for segmenting teeth in ct image Download PDF

Info

Publication number
WO2022108082A1
WO2022108082A1 PCT/KR2021/012952 KR2021012952W WO2022108082A1 WO 2022108082 A1 WO2022108082 A1 WO 2022108082A1 KR 2021012952 W KR2021012952 W KR 2021012952W WO 2022108082 A1 WO2022108082 A1 WO 2022108082A1
Authority
WO
WIPO (PCT)
Prior art keywords
tooth
mesh data
individual teeth
data
segmentation
Prior art date
Application number
PCT/KR2021/012952
Other languages
French (fr)
Korean (ko)
Inventor
김규년
김태형
Original Assignee
주식회사 쓰리디산업영상
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 주식회사 쓰리디산업영상 filed Critical 주식회사 쓰리디산업영상
Publication of WO2022108082A1 publication Critical patent/WO2022108082A1/en

Links

Images

Classifications

    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B6/00—Apparatus or devices for radiation diagnosis; Apparatus or devices for radiation diagnosis combined with radiation therapy equipment
    • A61B6/52—Devices using data or image processing specially adapted for radiation diagnosis
    • A61B6/5211—Devices using data or image processing specially adapted for radiation diagnosis involving processing of medical diagnostic data
    • A61B6/5217—Devices using data or image processing specially adapted for radiation diagnosis involving processing of medical diagnostic data extracting a diagnostic or physiological parameter from medical diagnostic data
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B6/00—Apparatus or devices for radiation diagnosis; Apparatus or devices for radiation diagnosis combined with radiation therapy equipment
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B6/00—Apparatus or devices for radiation diagnosis; Apparatus or devices for radiation diagnosis combined with radiation therapy equipment
    • A61B6/02—Arrangements for diagnosis sequentially in different planes; Stereoscopic radiation diagnosis
    • A61B6/03—Computed tomography [CT]
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B6/00—Apparatus or devices for radiation diagnosis; Apparatus or devices for radiation diagnosis combined with radiation therapy equipment
    • A61B6/02—Arrangements for diagnosis sequentially in different planes; Stereoscopic radiation diagnosis
    • A61B6/03—Computed tomography [CT]
    • A61B6/032—Transmission computed tomography [CT]
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B6/00—Apparatus or devices for radiation diagnosis; Apparatus or devices for radiation diagnosis combined with radiation therapy equipment
    • A61B6/50—Apparatus or devices for radiation diagnosis; Apparatus or devices for radiation diagnosis combined with radiation therapy equipment specially adapted for specific body parts; specially adapted for specific clinical applications
    • A61B6/51—Apparatus or devices for radiation diagnosis; Apparatus or devices for radiation diagnosis combined with radiation therapy equipment specially adapted for specific body parts; specially adapted for specific clinical applications for dentistry
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G06T7/0012—Biomedical image inspection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H30/00—ICT specially adapted for the handling or processing of medical images
    • G16H30/40—ICT specially adapted for the handling or processing of medical images for processing medical images, e.g. editing
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10072—Tomographic images
    • G06T2207/10081—Computed x-ray tomography [CT]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20081—Training; Learning
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20084—Artificial neural networks [ANN]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30004—Biomedical image processing
    • G06T2207/30036—Dental; Teeth

Definitions

  • the present disclosure (Disclosure) relates to a tooth division system and method as a whole, and more particularly, to a tooth division system and method for dividing a tooth from a dental CT image.
  • FIG. 1 is a view showing a flowchart of individual steps of the prior art (Republic of Korea Patent No. 10-1862378, 2018.05.23), the prior art relates to a system for automatically recognizing teeth on a CT-photographed oral image, CT An input unit 100 for receiving a photographed oral image, a preprocessing unit 200 for binarizing the oral image, and a parabola extending along the tooth arrangement on the binarized oral image, and tooth portion on the oral image
  • a system including a tooth scan unit 300 for recognizing each tooth individually by differentiating and judging a point perpendicular to the parabola as a boundary dividing the teeth is proposed.
  • the prior art relates to a system for automatically individually recognizing teeth in a CT-imaged oral image, extracting a tooth part by binarizing the CT image, and polynomial fitting the tooth area to extend a parabola along the tooth. was extracted and a method of segmenting adjacent teeth to be perpendicular to the parabola was proposed.
  • CT includes a preprocessor for thresholding tooth segmentation in the oral image, but the tooth crown portion appears as a larger value than the jawbone in the CT image, but the tooth root portion appears as the same or smaller value than the adjacent jawbone.
  • the method presented in the prior art has a problem in that segmentation is not possible up to the root portion of the tooth.
  • the parabola set by the tooth scan unit 300 is formed by polynomial fitting of the tooth part in the binarized oral image, but the arch shape is not a round shape, but a square or pointed shape depending on race. Since there is also the arch of the prior art, there is a problem that the versatility is poor.
  • the method according to the prior art does not provide numerical information of individual teeth.
  • an object of the present invention is to provide a tooth segmentation system and method capable of extracting not only the crown of the tooth but also the root region.
  • an input unit for receiving a CT-photographed oral image as first volume data; a tooth region extractor for extracting a tooth region from the first volume data input to the input unit using a semantic segmentation method, and generating first mesh data for the extracted tooth region; an adjacent tooth dividing unit generating second mesh data for individual teeth from the first mesh data to divide the individual teeth; and an output unit for outputting second mesh data for individual teeth.
  • a tooth segmentation system and method according to an aspect of the present invention can segment a tooth by extracting not only the crown of the tooth but also the root region.
  • FIG. 1 is a diagram showing a flowchart of individual steps in the prior art.
  • FIG. 2 is a diagram schematically showing the configuration of a tooth division system according to an embodiment of the present invention.
  • FIG. 3 is a flowchart illustrating a process performed in the tooth region extraction unit of the tooth division system according to an embodiment of the present invention.
  • FIG. 4 is a view schematically showing a tooth area extracted by the tooth area extraction unit of the tooth division system according to an embodiment of the present invention.
  • FIG. 5 is a diagram schematically illustrating a tooth area represented by first mesh data generated by a tooth area extraction unit of a tooth division system according to an embodiment of the present invention.
  • FIG. 6 is a flowchart illustrating a process performed in an adjacent tooth division of a tooth division system according to an embodiment of the present invention.
  • FIG. 7 is a diagram schematically showing individual teeth separated as second mesh data by adjacent tooth divisions according to an embodiment of the present invention.
  • FIG. 8 is a diagram schematically illustrating an individual tooth separated as second volume data by an adjacent tooth segment according to an embodiment of the present invention.
  • the tooth segmentation system and method according to an aspect of the present invention can be implemented by including the function in an application program installed and executed in a computer device or medical image analysis equipment.
  • Semantic segmentation is a technique for classifying each pixel of an image, and aims to segment the position of a specific object in the image by pixel unit. Semantic segmentation can segment an image more precisely than detection, which seeks to know whether a specific object exists in the image, or localization, which seeks to find out the coordinates of a specific object's location in the image. have.
  • As the method of semantic segmentation there are both methods using traditional machine learning such as Markov Random Fields and methods using neural networks such as Fully Convolution Network (FCN) and U-net. Deeplab V3+ announced by Google is the announced method with the best performance.
  • semantic segmentation is used in the step of extracting the tooth area from the 3D CT data of the tooth area extraction unit.
  • 3D mesh processing is a technology that performs various calculations by representing the surface of a 3D shape in a mesh form, and can dramatically reduce the data size and amount of calculations compared to expressing the 3D shape as a volume.
  • Mesh processing technology includes calculations such as mesh boolean algebra, plane slice, and mesh flattening, and is applied to computer graphics including games and CAD.
  • mesh cutting and flattening techniques are applied to segment adjacent teeth in the mesh extracted by the neural network.
  • FIG. 2 is a diagram schematically showing the configuration of a tooth division system according to an embodiment of the present invention.
  • the present invention is an input unit 1000 that receives a CT-photographed oral image as first volume data, and semantic segmentation from the first volume data input to the input unit 1000
  • the tooth region extractor 2000 extracts a tooth region using the method and generates first mesh data for the extracted tooth region, and generates second mesh data for individual teeth from the first mesh data to divide the individual teeth It is possible to provide a tooth segmentation system, including an adjacent tooth segmentation unit 3000 , and an output unit 4000 outputting second mesh data for individual teeth.
  • the tooth segmentation system may include an input unit 1000 that receives a CT-photographed oral image as first volume data.
  • the input unit 1000 receives a CT image of the oral cavity centered on the patient's teeth, and the oral image may be input to the input unit 1000 in the form of first volume data.
  • the input unit 1000 may target the entire oral CT image, or may set a region of interest (ROI) in the oral CT image to target only a portion.
  • ROI region of interest
  • the tooth segmentation system may include a tooth region extraction unit 2000 .
  • the tooth region extraction unit 2000 may extract a tooth region from the first volume data input to the input unit 1000 using a semantic segmentation method and generate first mesh data for the extracted tooth region. have.
  • FIG. 3 is a flowchart illustrating a process performed by the tooth region extraction unit 2000 of the tooth division system according to an embodiment of the present invention.
  • the tooth region extractor 2000 may receive first volume data of an oral image taken by CT as an input and may output first mesh data for a tooth region.
  • the tooth region extractor 2000 extracts the tooth region from all tomography layers of the CT image by using the semantic segmentation method by inputting the first volume data of the CT image of the oral cavity. can do.
  • the result extracted using the semantic segmentation method is an image that has the same size (ie, horizontal, vertical, and height) as the input CT image of the oral cavity. area can be distinguished.
  • the semantic segmentation used in the tooth region extraction unit 2000 is selected from, for example, a Markov Random Fields method, a Fully Convolution network (FCN) method, a U-net method, and a Deeplab V3+ method. may be any one of
  • FIG. 4 is a view schematically showing a tooth area extracted by the tooth area extraction unit 2000 of the tooth division system according to an embodiment of the present invention.
  • FIG. 4 is a photograph showing two tomographic layers and a tooth area corresponding to the two tomographic layers in the CT taken oral image of the input unit 1000, and (a) a tooth root portion and (b) a tooth crown portion among the extracted tooth areas. indicates The red area is the extracted tooth area. Referring to FIG. 4 , it can be confirmed that the tooth region extraction was well performed for both the root portion of the tooth and the crown portion of the tooth.
  • the tooth region correction is performed on a tomography of the first volume data, and a tomographic image parallel to the XY YZ XZ plane may be utilized. Since there may be discontinuities in the segmentation map for each tomography, the 3D Gaussian filter may be applied to the tooth region extracted from the first volume data before generating the first mesh data to make the surface smooth.
  • the tooth region extractor 2000 may generate first mesh data for the extracted tooth region using a semantic segmentation method.
  • the first mesh data generation algorithm may use, for example, a marching cubes algorithm, and if necessary, refining the first mesh data.
  • FIG. 5 is a diagram schematically illustrating a tooth area represented by the first mesh data generated by the tooth area extraction unit 2000 of the tooth division system according to an embodiment of the present invention. Although only the mandible is shown in FIG. 5, both upper and lower mesh data are generated. Such first mesh data is a result of the tooth region extraction unit 2000 in the tooth division system according to the present invention.
  • a tooth division system may include an adjacent tooth division 3000 .
  • FIG. 6 is a flowchart illustrating a process performed in the adjacent tooth division part 3000 of the tooth division system according to an embodiment of the present invention.
  • the adjacent tooth dividing unit 3000 may divide the individual teeth by generating second mesh data for the individual teeth from the first mesh data.
  • the adjacent tooth division unit 3000 receives an arbitrary point on each individual tooth to be divided in the tooth region of the first mesh data, and designates this point as a source or terminal And, by applying a graph segmentation algorithm to derive geodesic lines dividing each tooth, it is possible to generate second mesh data for individual teeth from the first mesh data. At this time, if the geodesic line is not drawn correctly, the user can correct the cut geodesic line by manually moving the point on the geodesic line.
  • the process of receiving an arbitrary point on each individual tooth to be divided in the tooth region of the first mesh data may be performed by the user clicking a point on each tooth to be divided in the first mesh data.
  • FIG. 7 is a diagram schematically showing individual teeth separated as second mesh data by the adjacent tooth division part 3000 according to an embodiment of the present invention. Color is assigned to distinguish teeth
  • the adjacent tooth dividing unit 3000 may divide the individual teeth by converting the second mesh data generated for the individual teeth from the first mesh data into the second volume data for the individual teeth.
  • the adjacent tooth division unit 3000 converts the second mesh data into the second volume data for the individual teeth
  • the tooth axis of the individual teeth and the tooth bounding are performed with respect to the second mesh data. It can be performed by deriving numerical information about the box and size.
  • the adjacent tooth dividing unit 3000 may convert the second mesh data into 3D second volume data composed of points included in the second mesh data.
  • the tooth axis and bounding box derivation is based on a clinical method, and in this case, information about the tooth size can be calculated from the 'width x length x height' of the bounding box.
  • FIG. 8 is a diagram schematically illustrating individual teeth separated as second volume data by an adjacent tooth division unit 3000 according to an embodiment of the present invention. Color is assigned to distinguish teeth
  • the tooth segmentation system may include an output unit 4000 .
  • the output unit 4000 may output the second mesh data for the individual teeth when the adjacent tooth division unit 3000 divides the individual teeth by generating the second mesh data for the individual teeth from the first mesh data. have.
  • the output unit 4000 converts the second mesh data generated for the individual teeth from the first mesh data into the second volume data for the individual teeth
  • the adjacent tooth division unit 3000 divides the individual teeth
  • the second volume data for individual teeth may be output.
  • the step of segmenting adjacent teeth comprises dividing the individual teeth by converting the second mesh data generated for the individual teeth from the first mesh data into second volume data for the individual teeth, and , the output step may provide a tooth segmentation method of outputting the second volume data converted in the adjacent tooth segmentation step.
  • the semantic segmentation used in the tooth region extraction step is a Markov Random Fields method, a Fully Convolution network (FCN) method, a U-net method , and Deeplab V3+ method, which is any one selected from the method, may provide a tooth segmentation method.
  • FCN Fully Convolution network
  • U-net U-net
  • Deeplab V3+ method Deeplab V3+
  • the present invention may provide a tooth segmentation method, wherein the tooth region extraction step generates first mesh data using a marching cubes algorithm.
  • the step of extracting the tooth area is to smooth the surface by applying a three-dimensional Gaussian filter to the tooth area extracted from the first volume data before generating the first mesh data.
  • a tooth segmentation method may be provided.
  • generating second mesh data for individual teeth from the first mesh data in the step of dividing adjacent teeth includes each individual tooth to be divided in the tooth region of the first mesh data. It is possible to provide a tooth segmentation method, which is performed by receiving an arbitrary point on a tooth, designating this point as a source or a terminal, and applying a graph segmentation algorithm to derive a geodesic line dividing each tooth.
  • converting the second mesh data to the second volume data for individual teeth in the step of dividing adjacent teeth is a tooth axis for an individual tooth with respect to the second mesh data, tooth bounding It is possible to provide a tooth segmentation method, which is performed by deriving numerical information about a box and a size.
  • Input unit 1000 tooth region extraction unit 2000, adjacent tooth division unit 3000, output unit 4000
  • an input unit for receiving a CT image of the oral cavity as first volume data; a tooth region extractor for extracting a tooth region from the first volume data input to the input unit using a semantic segmentation method, and generating first mesh data for the extracted tooth region; an adjacent tooth dividing unit generating second mesh data for individual teeth from the first mesh data to divide the individual teeth; and an output unit for outputting second mesh data for individual teeth.
  • the adjacent tooth division unit converts the second mesh data generated for the individual teeth from the first mesh data into the second volume data for the individual teeth to segment the individual teeth
  • the output unit converts the second mesh data generated for the individual teeth from the first mesh data into the second volume data for the individual teeth 2 Outputting volume data, tooth segmentation system.
  • the semantic segmentation used in the tooth region extraction unit is any one selected from the Markov Random Fields method, the Fully Convolution network (FCN) method, the U-net method, and the Deeplab V3+ method. Phosphorus, tooth segmentation system.
  • the tooth region extraction unit applies a three-dimensional Gaussian filter to the tooth region extracted from the first volume data before generating the first mesh data to smooth the surface.
  • the adjacent tooth division unit converts the second mesh data to the second volume data for the individual teeth by deriving numerical information about the tooth axis, the tooth bounding box, and the size of the individual teeth with respect to the second mesh data. Performed, tooth segmentation system.
  • the adjacent tooth segmentation step converts the second mesh data generated for the individual teeth from the first mesh data into the second volume data for the individual teeth to segment the individual teeth, and the output step is converted in the adjacent tooth segmentation step A tooth segmentation method of outputting the second volume data.
  • the semantic segmentation used in the tooth region extraction step is any one selected from the Markov Random Fields method, the Fully Convolution network (FCN) method, the U-net method, and the Deeplab V3+ method. Phosphorus, how to split teeth.
  • a three-dimensional Gaussian filter is applied to the tooth region extracted from the first volume data to smooth the surface.

Landscapes

  • Health & Medical Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Medical Informatics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Public Health (AREA)
  • General Health & Medical Sciences (AREA)
  • Physics & Mathematics (AREA)
  • Biomedical Technology (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Radiology & Medical Imaging (AREA)
  • Pathology (AREA)
  • Surgery (AREA)
  • Veterinary Medicine (AREA)
  • High Energy & Nuclear Physics (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Molecular Biology (AREA)
  • Biophysics (AREA)
  • Animal Behavior & Ethology (AREA)
  • Optics & Photonics (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Physics & Mathematics (AREA)
  • Epidemiology (AREA)
  • Primary Health Care (AREA)
  • Data Mining & Analysis (AREA)
  • Databases & Information Systems (AREA)
  • Dentistry (AREA)
  • Oral & Maxillofacial Surgery (AREA)
  • Physiology (AREA)
  • Pulmonology (AREA)
  • Quality & Reliability (AREA)
  • Apparatus For Radiation Diagnosis (AREA)

Abstract

The present disclosure relates overall to a system and a method for segmenting teeth, and more specifically relates to a system and a method for segmenting teeth which segment teeth from a dental CT image. The system and method for segmenting teeth, according to one embodiment of the present invention, can segment teeth by extracting not only the crown portion of a tooth but also extracting as far as the dental root area.

Description

CT 영상에서의 치아 분할 시스템 및 방법Tooth segmentation system and method in CT image
본 개시(Disclosure)는 전체적으로 치아 분할 시스템 및 방법에 관한 것으로, 더욱 상세하게는 치과용 CT 영상으로부터 치아를 분할하는 치아 분할 시스템 및 방법에 관한 것이다.The present disclosure (Disclosure) relates to a tooth division system and method as a whole, and more particularly, to a tooth division system and method for dividing a tooth from a dental CT image.
여기서는, 본 개시에 관한 배경기술이 제공되며, 이들이 반드시 공지기술을 의미하는 것은 아니다(This section provides background information related to the present disclosure which is not necessarily prior art).Herein, background information related to the present disclosure is provided, and they do not necessarily mean prior art (This section provides background information related to the present disclosure which is not necessarily prior art).
도 1은 종래기술(대한민국 등록특허 제10-1862378호, 2018.05.23)의 개별 단계 순서도를 나타낸 도면으로, 종래기술은 CT촬영된 구강 이미지상의 치아를 자동으로 개별 인식하는 시스템에 관한 것으로, CT촬영된 구강 이미지를 입력받는 입력부(100)와, 구강 이미지를 이진화(thresholding)시키는 전처리부(200)와, 이진화된 구강 이미지 상에 치아 배열을 따라 연장되는 포물선을 설정하고, 구강 이미지 상의 치아 부위를 미분하여 포물선과 수직되는 포인트를 치아를 구분하는 경계로 판단하여 각 치아를 개별적으로 인식하는 치아 스캔부(300)를 포함하는 시스템을 제안하였다. 구체적으로 종래기술은 CT 촬영된 구강 이미지의 치아를 자동으로 개별 인식하는 시스템에 관한 것으로, CT 이미지를 이진화하여 치아 부분을 추출하고, 치아 영역을 다항식 피팅(polynomial fitting)하여 치아를 따라 연장되는 포물선을 추출하여, 그 포물선에 수직이 되도록 인접 치아를 분할하는 방법을 제안하였다.1 is a view showing a flowchart of individual steps of the prior art (Republic of Korea Patent No. 10-1862378, 2018.05.23), the prior art relates to a system for automatically recognizing teeth on a CT-photographed oral image, CT An input unit 100 for receiving a photographed oral image, a preprocessing unit 200 for binarizing the oral image, and a parabola extending along the tooth arrangement on the binarized oral image, and tooth portion on the oral image A system including a tooth scan unit 300 for recognizing each tooth individually by differentiating and judging a point perpendicular to the parabola as a boundary dividing the teeth is proposed. Specifically, the prior art relates to a system for automatically individually recognizing teeth in a CT-imaged oral image, extracting a tooth part by binarizing the CT image, and polynomial fitting the tooth area to extend a parabola along the tooth. was extracted and a method of segmenting adjacent teeth to be perpendicular to the parabola was proposed.
그러나 종래기술의 경우 CT에서 치아 세그멘테이션을 구강 이미지에서 이진화(thresholding)시키는 전처리부를 포함하고 있으나 치아 크라운 부분은 CT 영상에서 턱뼈보다 큰 값으로 나타나지만 치아 뿌리 부분은 인접한 턱뼈와 같거나 작은 값으로 나타나므로 종래기술에 제시된 방법으로는 치아 뿌리 부분까지 세그멘테이션이 가능하지 않은 문제점이 있다. 또한 종래기술에서는 치아 스캔부(300)가 설정한 포물선은 이진화된 구강 이미지 중 치아 부위를 다항식 피팅(polynomial fitting)하여 형성된다고 개시되어 있으나, 악궁 모양은 인종에 따라 둥근 모양이 아닌 사각형이나 뾰족한 모양의 악궁도 존재하므로 종래기술의 방법은 범용성이 떨어지는 문제점이 있다. 또한 종래기술에 따르는 방법에서는 개별 치아의 수치정보를 제공하지는 않는 문제점이 있다.However, in the case of the prior art, CT includes a preprocessor for thresholding tooth segmentation in the oral image, but the tooth crown portion appears as a larger value than the jawbone in the CT image, but the tooth root portion appears as the same or smaller value than the adjacent jawbone. The method presented in the prior art has a problem in that segmentation is not possible up to the root portion of the tooth. In addition, in the prior art, it is disclosed that the parabola set by the tooth scan unit 300 is formed by polynomial fitting of the tooth part in the binarized oral image, but the arch shape is not a round shape, but a square or pointed shape depending on race. Since there is also the arch of the prior art, there is a problem that the versatility is poor. In addition, there is a problem in that the method according to the prior art does not provide numerical information of individual teeth.
본 발명은 하나의 양상에서 치아의 크라운뿐만 아니라 치근 영역까지 추출 가능한 치아 분할 시스템 및 방법을 제공하는 것을 목적으로 한다.In one aspect, an object of the present invention is to provide a tooth segmentation system and method capable of extracting not only the crown of the tooth but also the root region.
본 발명은 하나의 양상에서, CT 촬영된 구강 이미지를 제1 볼륨 데이터로 입력받는 입력부; 입력부에 입력된 제1 볼륨 데이터로부터 의미론적 분할(Semantic Segmentation) 방법을 사용하여 치아 영역을 추출하고, 추출된 치아 영역에 대하여 제1 메쉬 데이터를 생성하는 치아 영역 추출부; 제1 메쉬 데이터로부터 개별 치아에 대한 제2 메쉬 데이터를 생성하여 개별 치아를 분할하는 인접 치아 분할부; 및 개별 치아에 대한 제2 메쉬 데이터를 출력하는 출력부;를 포함하는, 치아 분할 시스템을 제공할 수 있다.The present invention in one aspect, an input unit for receiving a CT-photographed oral image as first volume data; a tooth region extractor for extracting a tooth region from the first volume data input to the input unit using a semantic segmentation method, and generating first mesh data for the extracted tooth region; an adjacent tooth dividing unit generating second mesh data for individual teeth from the first mesh data to divide the individual teeth; and an output unit for outputting second mesh data for individual teeth.
본 발명의 하나의 양상에 따르는 치아 분할 시스템 및 방법은 치아의 크라운뿐만 아니라 치근 영역까지 추출하여 치아를 분할할 수 있다.A tooth segmentation system and method according to an aspect of the present invention can segment a tooth by extracting not only the crown of the tooth but also the root region.
도 1은 종래 기술의 개별 단계 순서도를 나타낸 도면이다.1 is a diagram showing a flowchart of individual steps in the prior art.
도 2는 본 발명의 하나의 구체예에 따르는 치아 분할 시스템의 구성도를 개략적으로 나타낸 도면이다.2 is a diagram schematically showing the configuration of a tooth division system according to an embodiment of the present invention.
도 3은 본 발명의 하나의 구체예에 따르는 치아 분할 시스템의 치아 영역 추출부에서 수행되는 과정을 나타낸 순서도이다.3 is a flowchart illustrating a process performed in the tooth region extraction unit of the tooth division system according to an embodiment of the present invention.
도 4는 본 발명의 하나의 구체예에 따르는 치아 분할 시스템의 치아 영역 추출부에서 추출된 치아 영역을 개략적으로 나타낸 도면이다.4 is a view schematically showing a tooth area extracted by the tooth area extraction unit of the tooth division system according to an embodiment of the present invention.
도 5는 본 발명의 하나의 구체예에 따르는 치아 분할 시스템의 치아 영역 추출부에서 생성된 제1 메쉬 데이터에 의해 표현된 치아 영역을 개략적으로 나타낸 도면이다.5 is a diagram schematically illustrating a tooth area represented by first mesh data generated by a tooth area extraction unit of a tooth division system according to an embodiment of the present invention.
도 6은 본 발명의 하나의 구체예에 따르는 치아 분할 시스템의 인접 치아 분할부에서 수행되는 과정을 나타낸 순서도이다.6 is a flowchart illustrating a process performed in an adjacent tooth division of a tooth division system according to an embodiment of the present invention.
도 7은 본 발명의 하나의 구체예에 따르는 인접 치아 분할부에 의해 제2 메쉬 데이터로써 분리된 개별 치아를 개략적으로 나타내는 도면이다.7 is a diagram schematically showing individual teeth separated as second mesh data by adjacent tooth divisions according to an embodiment of the present invention.
도 8은 본 발명의 하나의 구체예에 따르는 인접 치아 분할부에 의해 제2 볼륨 데이터로써 분리된 개별 치아를 개략적으로 나타내는 도면이다. 8 is a diagram schematically illustrating an individual tooth separated as second volume data by an adjacent tooth segment according to an embodiment of the present invention.
본 발명의 하나의 양상에 따른 치아 분할 시스템 및 방법은 컴퓨터 장치에 설치되고 실행되는 응용프로그램 또는 의료 영상 분석 장비에 그 기능이 포함되는 것으로 실시가 가능하다.The tooth segmentation system and method according to an aspect of the present invention can be implemented by including the function in an application program installed and executed in a computer device or medical image analysis equipment.
의미론적 분할(Semantic Segmentation)은 이미지의 각 픽셀을 분류하는 기술로서, 이미지에 특정 사물이 어느 위치에 포함되어 있는지 픽셀 단위로 분할하는 것을 목표로 한다. 의미론적 분할 (Semantic Segmentation)은 특정 사물의 이미지 상 존재 여부를 알고자 하는 검출(detection)이나, 특정 사물이 이미지에서 위치하는 좌표를 알아내고자 하는 로컬라이제이션(Localization) 보다 이미지를 더 세밀하게 분할할 수 있다. 의미론적 분할(Semantic Segmentation) 방법으로는, 마르코프 랜덤필드(Markov Random Fields)와 같은 전통적인 기계학습을 사용하는 방법과 FCN(Fully Convolution network), U-net 등의 뉴럴 네트워크를 사용한 방법 모두 존재하며, Google에서 발표한 Deeplab V3+가 가장 성능이 좋다고 발표된 방법이다. 본 발병에서는 치아영역 추출부의 3차원 CT 데이터에서 치아 영역을 추출하는 단계에서 의미론적 분할(Semantic Segmentation)을 사용한다.Semantic segmentation is a technique for classifying each pixel of an image, and aims to segment the position of a specific object in the image by pixel unit. Semantic segmentation can segment an image more precisely than detection, which seeks to know whether a specific object exists in the image, or localization, which seeks to find out the coordinates of a specific object's location in the image. have. As the method of semantic segmentation, there are both methods using traditional machine learning such as Markov Random Fields and methods using neural networks such as Fully Convolution Network (FCN) and U-net. Deeplab V3+ announced by Google is the announced method with the best performance. In the present disease, semantic segmentation is used in the step of extracting the tooth area from the 3D CT data of the tooth area extraction unit.
3차원 메쉬 프로세싱은 3차원 형상의 표면을 메쉬 형태로 나타내어 각종 연산을 수행하는 기술로서, 3차원 형상을 볼륨으로 표현하는 것보다 데이터 크기와 연산량을 획기적으로 줄일 수 있다. 메쉬 처리 기술은 메쉬 부울대수, 평면 슬라이스, 메쉬 평탄화 등의 연산을 포함하며, 게임이나 CAD 를 비롯한 컴퓨터 그래픽 전반에 응용된다. 본 발명에서는 뉴럴 네트워크로 추출된 메쉬에서 인접 치아를 분할하는데 메쉬 절단 및 평탄화 기술이 적용된다.3D mesh processing is a technology that performs various calculations by representing the surface of a 3D shape in a mesh form, and can dramatically reduce the data size and amount of calculations compared to expressing the 3D shape as a volume. Mesh processing technology includes calculations such as mesh boolean algebra, plane slice, and mesh flattening, and is applied to computer graphics including games and CAD. In the present invention, mesh cutting and flattening techniques are applied to segment adjacent teeth in the mesh extracted by the neural network.
도 2는 본 발명의 하나의 구체예에 따르는 치아 분할 시스템의 구성도를 개략적으로 나타낸 도면이다.2 is a diagram schematically showing the configuration of a tooth division system according to an embodiment of the present invention.
본 발명은 하나의 양상의 하나의 구체예에서, CT 촬영된 구강 이미지를 제1 볼륨 데이터로 입력받는 입력부(1000), 입력부(1000)에 입력된 제1 볼륨 데이터로부터 의미론적 분할(Semantic Segmentation) 방법을 사용하여 치아 영역을 추출하고 추출된 치아 영역에 대하여 제1 메쉬 데이터를 생성하는 치아 영역 추출부(2000), 제1 메쉬 데이터로부터 개별 치아에 대한 제2 메쉬 데이터를 생성하여 개별 치아를 분할하는 인접 치아 분할부(3000), 및 개별 치아에 대한 제2 메쉬 데이터를 출력하는 출력부(4000)를 포함하는, 치아 분할 시스템을 제공할 수 있다.In one embodiment of one aspect, the present invention is an input unit 1000 that receives a CT-photographed oral image as first volume data, and semantic segmentation from the first volume data input to the input unit 1000 The tooth region extractor 2000 extracts a tooth region using the method and generates first mesh data for the extracted tooth region, and generates second mesh data for individual teeth from the first mesh data to divide the individual teeth It is possible to provide a tooth segmentation system, including an adjacent tooth segmentation unit 3000 , and an output unit 4000 outputting second mesh data for individual teeth.
구체적으로, 도 2를 참조하면, 본 발명의 하나의 양상의 하나의 구체예에 따른 치아 분할 시스템은 CT 촬영된 구강 이미지를 제1 볼륨 데이터로 입력받는 입력부(1000)를 포함할 수 있다.Specifically, referring to FIG. 2 , the tooth segmentation system according to an embodiment of an aspect of the present invention may include an input unit 1000 that receives a CT-photographed oral image as first volume data.
입력부(1000)는 환자의 치아를 중심으로 CT 촬영된 구강 이미지를 입력받는데, 이러한 구강 이미지는 제1 볼륨 데이터 형태로 입력부(1000)에 입력될 수 있다.The input unit 1000 receives a CT image of the oral cavity centered on the patient's teeth, and the oral image may be input to the input unit 1000 in the form of first volume data.
입력부(1000)는 구강 CT 촬영 이미지 전체를 대상으로 할 수도 있고, 구강 CT 촬영 이미지 중에서 관심 영역(Region of Interest, ROI)을 설정하여 일부만 대상으로 할 수도 있다.The input unit 1000 may target the entire oral CT image, or may set a region of interest (ROI) in the oral CT image to target only a portion.
또한 본 발명의 하나의 양상의 하나의 구체예에 따른 치아 분할 시스템은 치아 영역 추출부(2000)를 포함할 수 있다.Also, the tooth segmentation system according to one embodiment of one aspect of the present invention may include a tooth region extraction unit 2000 .
치아 영역 추출부(2000)는 입력부(1000)에 입력된 제1 볼륨 데이터로부터 의미론적 분할(Semantic Segmentation) 방법을 사용하여 치아 영역을 추출하고 추출된 치아 영역에 대하여 제1 메쉬 데이터를 생성할 수 있다.The tooth region extraction unit 2000 may extract a tooth region from the first volume data input to the input unit 1000 using a semantic segmentation method and generate first mesh data for the extracted tooth region. have.
도 3은 본 발명의 하나의 구체예에 따르는 치아 분할 시스템의 치아 영역 추출부(2000)에서 수행되는 과정을 나타낸 순서도이다.3 is a flowchart illustrating a process performed by the tooth region extraction unit 2000 of the tooth division system according to an embodiment of the present invention.
도 3을 참조하여 구체적으로 설명하면, 치아 영역 추출부(2000)는 CT 촬영된 구강 이미지의 제1 볼륨 데이터를 입력으로 하고 치아 영역에 대한 제1 메쉬 데이터를 출력으로 할 수 있다. Referring to FIG. 3 , the tooth region extractor 2000 may receive first volume data of an oral image taken by CT as an input and may output first mesh data for a tooth region.
더욱 구체적으로 설명하면, 먼저 치아 영역 추출부(2000)는 CT 촬영된 구강 이미지의 제1 볼륨 데이터를 입력으로 하여 의미론적 분할(Semantic Segmentation) 방법을 사용하여 CT 영상의 모든 단층에서 치아 영역을 추출할 수 있다. 의미론적 분할(Semantic Segmentation) 방법을 사용하여 추출된 결과는 입력한 CT 촬영된 구강 이미지와 크기(즉, 가로, 세로, 높이)가 동일한 영상으로서, 영상의 픽셀이나 복셀을 치아 영역과 치아가 아닌 영역으로 구별할 수 있다.More specifically, first, the tooth region extractor 2000 extracts the tooth region from all tomography layers of the CT image by using the semantic segmentation method by inputting the first volume data of the CT image of the oral cavity. can do. The result extracted using the semantic segmentation method is an image that has the same size (ie, horizontal, vertical, and height) as the input CT image of the oral cavity. area can be distinguished.
치아 영역 추출부(2000)에서 사용되는 의미론적 분할(Semantic Segmentation)은 예를 들어 마르코프 랜덤필드(Markov Random Fields) 방법, FCN(Fully Convolution network) 방법, U-net 방법, 및 Deeplab V3+ 방법 중에서 선택되는 어느 하나일 수 있다.The semantic segmentation used in the tooth region extraction unit 2000 is selected from, for example, a Markov Random Fields method, a Fully Convolution network (FCN) method, a U-net method, and a Deeplab V3+ method. may be any one of
도 4는 본 발명의 하나의 구체예에 따르는 치아 분할 시스템의 치아 영역 추출부(2000)에서 추출된 치아 영역을 개략적으로 나타낸 도면이다. 4 is a view schematically showing a tooth area extracted by the tooth area extraction unit 2000 of the tooth division system according to an embodiment of the present invention.
구체적으로 도 4는 입력부(1000)의 CT 촬영된 구강 이미지 중 두 개의 단층과 이것에 대응하는 치아 영역을 겹쳐서 나타낸 사진으로, 추출된 치아 영역 중 (a) 치아 뿌리 부분과 (b) 치아 크라운 부분을 나타낸다. 붉은 영역이 추출된 치아 영역이다. 도 4에 의하면 치아의 뿌리 부분과 치아의 크라운 부분 모두 치아 영역 추출이 잘 이루어졌음을 확인할 수 있다.Specifically, FIG. 4 is a photograph showing two tomographic layers and a tooth area corresponding to the two tomographic layers in the CT taken oral image of the input unit 1000, and (a) a tooth root portion and (b) a tooth crown portion among the extracted tooth areas. indicates The red area is the extracted tooth area. Referring to FIG. 4 , it can be confirmed that the tooth region extraction was well performed for both the root portion of the tooth and the crown portion of the tooth.
한편, 만약 도 4와 같은 출력에서 부정확한 부분이 발견되는 경우, 수동으로 이를 수정할 수 있다. 치아 영역 수정은 제1 볼륨 데이터의 단층에서 이루어지는데, XY YZ XZ 평면과 평행한 단층 이미지가 활용될 수 있다. 각 단층별로 세그멘테이션 맵(Segmentation map)에 불연속성이 있을 수 있으므로, 제1 메쉬 데이터를 생성하기 이전에 제1 볼륨 데이터로부터 추출된 치아 영역에 대하여 3차원 가우시안 필터를 적용하여 표면을 매끄럽게 할 수 있다.Meanwhile, if an inaccurate part is found in the output as shown in FIG. 4 , it can be manually corrected. The tooth region correction is performed on a tomography of the first volume data, and a tomographic image parallel to the XY YZ XZ plane may be utilized. Since there may be discontinuities in the segmentation map for each tomography, the 3D Gaussian filter may be applied to the tooth region extracted from the first volume data before generating the first mesh data to make the surface smooth.
이후 치아 영역 추출부(2000)는 의미론적 분할(Semantic Segmentation) 방법을 사용하여 추출된 치아 영역에 대하여 제1 메쉬 데이터를 생성할 수 있다. 제1 메쉬 데이터 생성 알고리즘은 예컨대 마칭 큐브(marching cubes) 알고리즘을 사용할 수 있으며, 필요한 경우 제1 메쉬 데이터를 재정의(refining)할 수도 있다.Thereafter, the tooth region extractor 2000 may generate first mesh data for the extracted tooth region using a semantic segmentation method. The first mesh data generation algorithm may use, for example, a marching cubes algorithm, and if necessary, refining the first mesh data.
도 5는 본 발명의 하나의 구체예에 따르는 치아 분할 시스템의 치아 영역 추출부(2000)에서 생성된 제1 메쉬 데이터에 의해 표현된 치아 영역을 개략적으로 나타낸 도면이다. 도 5에는 하악만 제시되어 있으나 상악과 하악 메쉬 데이터가 모두 생성된다. 이와 같은 제1 메쉬 데이터가 본 발명에 따른 치아 분할 시스템에서 치아 영역 추출부(2000)의 결과이다.5 is a diagram schematically illustrating a tooth area represented by the first mesh data generated by the tooth area extraction unit 2000 of the tooth division system according to an embodiment of the present invention. Although only the mandible is shown in FIG. 5, both upper and lower mesh data are generated. Such first mesh data is a result of the tooth region extraction unit 2000 in the tooth division system according to the present invention.
또한 본 발명의 하나의 양상의 하나의 구체예에 따른 치아 분할 시스템은 인접 치아 분할부(3000)를 포함할 수 있다.Also, a tooth division system according to one embodiment of an aspect of the present invention may include an adjacent tooth division 3000 .
도 6은 본 발명의 하나의 구체예에 따르는 치아 분할 시스템의 인접 치아 분할부(3000)에서 수행되는 과정을 나타낸 순서도이다.6 is a flowchart illustrating a process performed in the adjacent tooth division part 3000 of the tooth division system according to an embodiment of the present invention.
본 발명의 하나의 구체예에서 인접 치아 분할부(3000)는 제1 메쉬 데이터로부터 개별 치아에 대한 제2 메쉬 데이터를 생성하여 개별 치아를 분할할 수 있다.In one embodiment of the present invention, the adjacent tooth dividing unit 3000 may divide the individual teeth by generating second mesh data for the individual teeth from the first mesh data.
도 6을 참조하여 구체적으로 설명하면, 인접 치아 분할부(3000)는 제1 메쉬 데이터의 치아 영역에서 분할하고자 하는 각각의 개별 치아 위에 임의의 한 점을 입력받고, 이러한 점을 소스 또는 터미널로 지정하고, 그래프 분할 알고리즘을 적용하여 각 치아를 분할하는 측지선을 도출함으로써, 제1 메쉬 데이터로부터 개별 치아에 대한 제2 메쉬 데이터를 생성할 수 있다. 이때 측지선이 정확히 그려지지 않았다면 사용자는 측지선 위의 점을 수동으로 이동함으로써 절단 측지선을 수정할 수 있다. 또한 제1 메쉬 데이터의 치아 영역에서 분할하고자 하는 각각의 개별 치아 위에 임의의 한 점을 입력받는 과정은 사용자가 제1 메쉬 데이터에서 분할하고자 하는 각각의 치아 위에 한 점을 클릭함으로써 이루어질 수 있다.6, the adjacent tooth division unit 3000 receives an arbitrary point on each individual tooth to be divided in the tooth region of the first mesh data, and designates this point as a source or terminal And, by applying a graph segmentation algorithm to derive geodesic lines dividing each tooth, it is possible to generate second mesh data for individual teeth from the first mesh data. At this time, if the geodesic line is not drawn correctly, the user can correct the cut geodesic line by manually moving the point on the geodesic line. In addition, the process of receiving an arbitrary point on each individual tooth to be divided in the tooth region of the first mesh data may be performed by the user clicking a point on each tooth to be divided in the first mesh data.
도 7은 본 발명의 하나의 구체예에 따르는 인접 치아 분할부(3000)에 의해 제2 메쉬 데이터로써 분리된 개별 치아를 개략적으로 나타내는 도면이다. 색상은 치아의 구분을 위해 지정한 것이다7 is a diagram schematically showing individual teeth separated as second mesh data by the adjacent tooth division part 3000 according to an embodiment of the present invention. Color is assigned to distinguish teeth
한편 인접 치아 분할부(3000)는 제1 메쉬 데이터로부터 개별 치아에 대하여 생성된 제2 메쉬 데이터를 개별 치아에 대한 제2 볼륨 데이터로 변환하여 개별 치아를 분할할 수 있다.Meanwhile, the adjacent tooth dividing unit 3000 may divide the individual teeth by converting the second mesh data generated for the individual teeth from the first mesh data into the second volume data for the individual teeth.
도 6을 참조하여 구체적으로 설명하면, 인접 치아 분할부(3000)가 제2 메쉬 데이터를 개별 치아에 대한 제2 볼륨 데이터로 변환하는 것은 제2 메쉬 데이터에 대하여 개별 치아에 대한 치아축, 치아 바운딩박스, 및 크기에 대한 수치정보를 도출함으로써 수행될 수 있다.Referring specifically to FIG. 6 , when the adjacent tooth division unit 3000 converts the second mesh data into the second volume data for the individual teeth, the tooth axis of the individual teeth and the tooth bounding are performed with respect to the second mesh data. It can be performed by deriving numerical information about the box and size.
즉 인접 치아 분할부(3000)는 제2 메쉬 데이터를 제2 메쉬 데이터 내에 포함된 점들로 구성된 3차원 제2 볼륨 데이터로 변환할 수 있다. 치아의 축과 바운딩 박스 도출은 임상적 방법을 근거로 하며, 이때 치아 크기에 관한 정보는 바운딩 박스의 '가로 x 세로 x 높이'로부터 계산할 수 있다.That is, the adjacent tooth dividing unit 3000 may convert the second mesh data into 3D second volume data composed of points included in the second mesh data. The tooth axis and bounding box derivation is based on a clinical method, and in this case, information about the tooth size can be calculated from the 'width x length x height' of the bounding box.
도 8은 본 발명의 하나의 구체예에 따르는 인접 치아 분할부(3000)에 의해 제2 볼륨 데이터로써 분리된 개별 치아를 개략적으로 나타내는 도면이다. 색상은 치아의 구분을 위해 지정한 것이다8 is a diagram schematically illustrating individual teeth separated as second volume data by an adjacent tooth division unit 3000 according to an embodiment of the present invention. Color is assigned to distinguish teeth
또한 본 발명의 하나의 양상의 하나의 구체예에 따른 치아 분할 시스템은 출력부(4000)를 포함할 수 있다.Also, the tooth segmentation system according to one embodiment of one aspect of the present invention may include an output unit 4000 .
출력부(4000)는, 인접 치아 분할부(3000)가 제1 메쉬 데이터로부터 개별 치아에 대한 제2 메쉬 데이터를 생성하여 개별 치아를 분할하는 경우, 개별 치아에 대한 제2 메쉬 데이터를 출력할 수 있다.The output unit 4000 may output the second mesh data for the individual teeth when the adjacent tooth division unit 3000 divides the individual teeth by generating the second mesh data for the individual teeth from the first mesh data. have.
또한 출력부(4000)는, 인접 치아 분할부(3000)가 제1 메쉬 데이터로부터 개별 치아에 대하여 생성된 제2 메쉬 데이터를 개별 치아에 대한 제2 볼륨 데이터로 변환하여 개별 치아를 분할하는 경우, 개별 치아에 대한 제2 볼륨 데이터를 출력할 수 있다.In addition, when the output unit 4000 converts the second mesh data generated for the individual teeth from the first mesh data into the second volume data for the individual teeth, the adjacent tooth division unit 3000 divides the individual teeth, The second volume data for individual teeth may be output.
본 발명은 또 다른 양상의 하나의 구체예에서, CT 촬영된 구강 이미지를 제1 볼륨 데이터로 입력받는 입력단계; 입력단계에서 입력된 제1 볼륨 데이터로부터 의미론적 분할(Semantic Segmentation) 방법을 사용하여 치아 영역을 추출하고, 추출된 치아 영역에 대하여 제1 메쉬 데이터를 생성하는 치아 영역 추출 단계; 제1 메쉬 데이터로부터 개별 치아에 대한 제2 메쉬 데이터를 생성하여 개별 치아를 분할하는 인접 치아 분할 단계; 및 개별 치아에 대한 제2 메쉬 데이터를 출력하는 출력 단계;를 포함하는, 치아 분할 방법을 제공할 수 있다.In one embodiment of another aspect, the present invention, an input step of receiving a CT-photographed oral image as first volume data; a tooth region extraction step of extracting a tooth region using a semantic segmentation method from the first volume data input in the input step, and generating first mesh data for the extracted tooth region; an adjacent tooth segmentation step of dividing the individual teeth by generating second mesh data for the individual teeth from the first mesh data; and an output step of outputting second mesh data for individual teeth.
본 발명은 또 다른 양상의 또 다른 구체예에서, 인접 치아 분할 단계는 제1 메쉬 데이터로부터 개별 치아에 대하여 생성된 제2 메쉬 데이터를 개별 치아에 대한 제2 볼륨 데이터로 변환하여 개별 치아를 분할하고, 출력 단계는 인접 치아 분할 단계에서 변환된 제2 볼륨 데이터를 출력하는 치아 분할 방법을 제공할 수 있다.In another embodiment of another aspect, the step of segmenting adjacent teeth comprises dividing the individual teeth by converting the second mesh data generated for the individual teeth from the first mesh data into second volume data for the individual teeth, and , the output step may provide a tooth segmentation method of outputting the second volume data converted in the adjacent tooth segmentation step.
본 발명은 또 다른 양상의 또 다른 구체예에서, 치아 영역 추출 단계에서 사용되는 의미론적 분할(Semantic Segmentation)은 마르코프 랜덤필드(Markov Random Fields) 방법, FCN(Fully Convolution network) 방법, U-net 방법, 및 Deeplab V3+ 방법 중에서 선택되는 어느 하나인, 치아 분할 방법을 제공할 수 있다.In another embodiment of another aspect, the semantic segmentation used in the tooth region extraction step is a Markov Random Fields method, a Fully Convolution network (FCN) method, a U-net method , and Deeplab V3+ method, which is any one selected from the method, may provide a tooth segmentation method.
본 발명은 또 다른 양상의 또 다른 구체예에서, 치아 영역 추출 단계는 마칭 큐브(marching cubes) 알고리즘을 사용하여 제1 메쉬 데이터를 생성하는, 치아 분할 방법을 제공할 수 있다.In yet another embodiment of another aspect, the present invention may provide a tooth segmentation method, wherein the tooth region extraction step generates first mesh data using a marching cubes algorithm.
본 발명은 또 다른 양상의 또 다른 구체예에서, 치아 영역 추출 단계는 제1 메쉬 데이터를 생성하기 이전에, 제1 볼륨 데이터로부터 추출된 치아 영역에 대하여 3차원 가우시안 필터를 적용하여 표면을 매끄럽게 하는, 치아 분할 방법을 제공할 수 있다.In another embodiment of the present invention, in another embodiment of another aspect, the step of extracting the tooth area is to smooth the surface by applying a three-dimensional Gaussian filter to the tooth area extracted from the first volume data before generating the first mesh data. , a tooth segmentation method may be provided.
본 발명은 또 다른 양상의 또 다른 구체예에서, 인접 치아 분할 단계에서 제1 메쉬 데이터로부터 개별 치아에 대한 제2 메쉬 데이터를 생성하는 것은, 제1 메쉬 데이터의 치아 영역에서 분할하고자 하는 각각의 개별 치아 위에 임의의 한 점을 입력받고, 이러한 점을 소스 또는 터미널로 지정하고, 그래프 분할 알고리즘을 적용하여 각 치아를 분할하는 측지선을 도출함으로써 수행되는, 치아 분할 방법을 제공할 수 있다.In another embodiment of another aspect of the present invention, generating second mesh data for individual teeth from the first mesh data in the step of dividing adjacent teeth includes each individual tooth to be divided in the tooth region of the first mesh data. It is possible to provide a tooth segmentation method, which is performed by receiving an arbitrary point on a tooth, designating this point as a source or a terminal, and applying a graph segmentation algorithm to derive a geodesic line dividing each tooth.
본 발명은 또 다른 양상의 또 다른 구체예에서, 인접 치아 분할 단계에서 제2 메쉬 데이터를 개별 치아에 대한 제2 볼륨 데이터로 변환하는 것은 제2 메쉬 데이터에 대하여 개별 치아에 대한 치아축, 치아 바운딩박스, 및 크기에 대한 수치정보를 도출함으로써 수행되는, 치아 분할 방법을 제공할 수 있다.In another embodiment of another aspect of the present invention, converting the second mesh data to the second volume data for individual teeth in the step of dividing adjacent teeth is a tooth axis for an individual tooth with respect to the second mesh data, tooth bounding It is possible to provide a tooth segmentation method, which is performed by deriving numerical information about a box and a size.
입력부(1000), 치아 영역 추출부(2000), 인접 치아 분할부(3000), 출력부(4000) Input unit 1000, tooth region extraction unit 2000, adjacent tooth division unit 3000, output unit 4000
이하 본 발명의 다양한 실시예를 설명한다.Hereinafter, various embodiments of the present invention will be described.
(1) CT 촬영된 구강 이미지를 제1 볼륨 데이터로 입력받는 입력부; 입력부에 입력된 제1 볼륨 데이터로부터 의미론적 분할(Semantic Segmentation) 방법을 사용하여 치아 영역을 추출하고, 추출된 치아 영역에 대하여 제1 메쉬 데이터를 생성하는 치아 영역 추출부; 제1 메쉬 데이터로부터 개별 치아에 대한 제2 메쉬 데이터를 생성하여 개별 치아를 분할하는 인접 치아 분할부; 및 개별 치아에 대한 제2 메쉬 데이터를 출력하는 출력부;를 포함하는, 치아 분할 시스템.(1) an input unit for receiving a CT image of the oral cavity as first volume data; a tooth region extractor for extracting a tooth region from the first volume data input to the input unit using a semantic segmentation method, and generating first mesh data for the extracted tooth region; an adjacent tooth dividing unit generating second mesh data for individual teeth from the first mesh data to divide the individual teeth; and an output unit for outputting second mesh data for individual teeth.
(2) 인접 치아 분할부는 제1 메쉬 데이터로부터 개별 치아에 대하여 생성된 제2 메쉬 데이터를 개별 치아에 대한 제2 볼륨 데이터로 변환하여 개별 치아를 분할하고, 출력부는 인접 치아 분할부에서 변환된 제2 볼륨 데이터를 출력하는, 치아 분할 시스템.(2) the adjacent tooth division unit converts the second mesh data generated for the individual teeth from the first mesh data into the second volume data for the individual teeth to segment the individual teeth, and the output unit converts the second mesh data generated for the individual teeth from the first mesh data into the second volume data for the individual teeth 2 Outputting volume data, tooth segmentation system.
(3) 치아 영역 추출부에서 사용되는 의미론적 분할(Semantic Segmentation)은 마르코프 랜덤필드(Markov Random Fields) 방법, FCN(Fully Convolution network) 방법, U-net 방법, 및 Deeplab V3+ 방법 중에서 선택되는 어느 하나인, 치아 분할 시스템.(3) The semantic segmentation used in the tooth region extraction unit is any one selected from the Markov Random Fields method, the Fully Convolution network (FCN) method, the U-net method, and the Deeplab V3+ method. Phosphorus, tooth segmentation system.
(4) 치아 영역 추출부는 마칭 큐브(marching cubes) 알고리즘을 사용하여 제1 메쉬 데이터를 생성하는, 치아 분할 시스템.(4) the tooth segmentation system, wherein the tooth region extraction unit generates the first mesh data using a marching cubes algorithm.
(5) 치아 영역 추출부는 제1 메쉬 데이터를 생성하기 이전에, 제1 볼륨 데이터로부터 추출된 치아 영역에 대하여 3차원 가우시안 필터를 적용하여 표면을 매끄럽게 하는, 치아 분할 시스템.(5) The tooth region extraction unit applies a three-dimensional Gaussian filter to the tooth region extracted from the first volume data before generating the first mesh data to smooth the surface.
(6) 인접 치아 분할부가 제1 메쉬 데이터로부터 개별 치아에 대한 제2 메쉬 데이터를 생성하는 것은, 제1 메쉬 데이터의 치아 영역에서 분할하고자 하는 각각의 개별 치아 위에 임의의 한 점을 입력받고, 이러한 점을 소스 또는 터미널로 지정하고, 그래프 분할 알고리즘을 적용하여 각 치아를 분할하는 측지선을 도출함으로써 수행되는, 치아 분할 시스템.(6) When the adjacent tooth division unit generates the second mesh data for the individual teeth from the first mesh data, an arbitrary point is input on each individual tooth to be divided in the tooth region of the first mesh data, and the A tooth segmentation system performed by specifying a point as a source or terminal, and applying a graph segmentation algorithm to derive geodesic lines segmenting each tooth.
(7) 인접 치아 분할부가 제2 메쉬 데이터를 개별 치아에 대한 제2 볼륨 데이터로 변환하는 것은 제2 메쉬 데이터에 대하여 개별 치아에 대한 치아축, 치아 바운딩박스, 및 크기에 대한 수치정보를 도출함으로써 수행되는, 치아 분할 시스템.(7) The adjacent tooth division unit converts the second mesh data to the second volume data for the individual teeth by deriving numerical information about the tooth axis, the tooth bounding box, and the size of the individual teeth with respect to the second mesh data. Performed, tooth segmentation system.
(8) CT 촬영된 구강 이미지를 제1 볼륨 데이터로 입력받는 입력단계; 입력단계에서 입력된 제1 볼륨 데이터로부터 의미론적 분할(Semantic Segmentation) 방법을 사용하여 치아 영역을 추출하고, 추출된 치아 영역에 대하여 제1 메쉬 데이터를 생성하는 치아 영역 추출 단계; 제1 메쉬 데이터로부터 개별 치아에 대한 제2 메쉬 데이터를 생성하여 개별 치아를 분할하는 인접 치아 분할 단계; 및 개별 치아에 대한 제2 메쉬 데이터를 출력하는 출력 단계;를 포함하는, 치아 분할 방법.(8) an input step of receiving a CT image of the oral cavity as first volume data; a tooth region extraction step of extracting a tooth region using a semantic segmentation method from the first volume data input in the input step, and generating first mesh data for the extracted tooth region; an adjacent tooth segmentation step of dividing the individual teeth by generating second mesh data for the individual teeth from the first mesh data; and an output step of outputting second mesh data for individual teeth.
(9) 인접 치아 분할 단계는 제1 메쉬 데이터로부터 개별 치아에 대하여 생성된 제2 메쉬 데이터를 개별 치아에 대한 제2 볼륨 데이터로 변환하여 개별 치아를 분할하고, 출력 단계는 인접 치아 분할 단계에서 변환된 제2 볼륨 데이터를 출력하는, 치아 분할 방법.(9) The adjacent tooth segmentation step converts the second mesh data generated for the individual teeth from the first mesh data into the second volume data for the individual teeth to segment the individual teeth, and the output step is converted in the adjacent tooth segmentation step A tooth segmentation method of outputting the second volume data.
(10) 치아 영역 추출 단계에서 사용되는 의미론적 분할(Semantic Segmentation)은 마르코프 랜덤필드(Markov Random Fields) 방법, FCN(Fully Convolution network) 방법, U-net 방법, 및 Deeplab V3+ 방법 중에서 선택되는 어느 하나인, 치아 분할 방법.(10) The semantic segmentation used in the tooth region extraction step is any one selected from the Markov Random Fields method, the Fully Convolution network (FCN) method, the U-net method, and the Deeplab V3+ method. Phosphorus, how to split teeth.
(11) 치아 영역 추출 단계는 마칭 큐브(marching cubes) 알고리즘을 사용하여 제1 메쉬 데이터를 생성하는, 치아 분할 방법.(11) The tooth segmentation method, wherein the tooth region extraction step generates first mesh data using a marching cubes algorithm.
(12) 치아 영역 추출 단계는 제1 메쉬 데이터를 생성하기 이전에, 제1 볼륨 데이터로부터 추출된 치아 영역에 대하여 3차원 가우시안 필터를 적용하여 표면을 매끄럽게 하는, 치아 분할 방법.(12) In the tooth region extraction step, before generating the first mesh data, a three-dimensional Gaussian filter is applied to the tooth region extracted from the first volume data to smooth the surface.
(13) 인접 치아 분할 단계에서 제1 메쉬 데이터로부터 개별 치아에 대한 제2 메쉬 데이터를 생성하는 것은, 제1 메쉬 데이터의 치아 영역에서 분할하고자 하는 각각의 개별 치아 위에 임의의 한 점을 입력받고, 이러한 점을 소스 또는 터미널로 지정하고, 그래프 분할 알고리즘을 적용하여 각 치아를 분할하는 측지선을 도출함으로써 수행되는, 치아 분할 방법.(13) In the step of dividing adjacent teeth, generating second mesh data for individual teeth from the first mesh data receives an arbitrary point on each individual tooth to be divided in the tooth area of the first mesh data, A tooth segmentation method performed by designating these points as a source or a terminal, and applying a graph segmentation algorithm to derive geodesic lines segmenting each tooth.
(14) 인접 치아 분할 단계에서 제2 메쉬 데이터를 개별 치아에 대한 제2 볼륨 데이터로 변환하는 것은 제2 메쉬 데이터에 대하여 개별 치아에 대한 치아축, 치아 바운딩박스, 및 크기에 대한 수치정보를 도출함으로써 수행되는, 치아 분할 방법.(14) In the step of dividing adjacent teeth, converting the second mesh data to the second volume data for individual teeth derives numerical information about the tooth axis, the tooth bounding box, and the size of the individual teeth with respect to the second mesh data A tooth segmentation method performed by

Claims (14)

  1. CT 촬영된 구강 이미지를 제1 볼륨 데이터로 입력받는 입력부;an input unit receiving a CT-photographed oral image as first volume data;
    입력부에 입력된 제1 볼륨 데이터로부터 의미론적 분할(Semantic Segmentation) 방법을 사용하여 치아 영역을 추출하고, 추출된 치아 영역에 대하여 제1 메쉬 데이터를 생성하는 치아 영역 추출부;a tooth region extractor for extracting a tooth region from the first volume data input to the input unit using a semantic segmentation method, and generating first mesh data for the extracted tooth region;
    제1 메쉬 데이터로부터 개별 치아에 대한 제2 메쉬 데이터를 생성하여 개별 치아를 분할하는 인접 치아 분할부; 및an adjacent tooth dividing unit generating second mesh data for individual teeth from the first mesh data to divide the individual teeth; and
    개별 치아에 대한 제2 메쉬 데이터를 출력하는 출력부;an output unit for outputting second mesh data for individual teeth;
    를 포함하는, 치아 분할 시스템.Including, tooth segmentation system.
  2. 청구항 1에 있어서,The method according to claim 1,
    인접 치아 분할부는 제1 메쉬 데이터로부터 개별 치아에 대하여 생성된 제2 메쉬 데이터를 개별 치아에 대한 제2 볼륨 데이터로 변환하여 개별 치아를 분할하고,The adjacent tooth division unit divides the individual teeth by converting the second mesh data generated for the individual teeth from the first mesh data into the second volume data for the individual teeth,
    출력부는 인접 치아 분할부에서 변환된 제2 볼륨 데이터를 출력하는, 치아 분할 시스템.The output unit outputs the second volume data converted by the adjacent tooth division unit, the tooth division system.
  3. 청구항 1 또는 청구항 2에 있어서,The method according to claim 1 or 2,
    치아 영역 추출부에서 사용되는 의미론적 분할(Semantic Segmentation)은 마르코프 랜덤필드(Markov Random Fields) 방법, FCN(Fully Convolution network) 방법, U-net 방법, 및 Deeplab V3+ 방법 중에서 선택되는 어느 하나인, 치아 분할 시스템.The semantic segmentation used in the tooth region extraction unit is any one selected from the Markov Random Fields method, the Fully Convolution network (FCN) method, the U-net method, and the Deeplab V3+ method. split system.
  4. 청구항 1 또는 청구항 2에 있어서,,The method according to claim 1 or 2,
    치아 영역 추출부는 마칭 큐브(marching cubes) 알고리즘을 사용하여 제1 메쉬 데이터를 생성하는, 치아 분할 시스템.The tooth segmentation system, wherein the tooth region extractor generates the first mesh data using a marching cubes algorithm.
  5. 청구항 1 또는 청구항 2에 있어서,The method according to claim 1 or 2,
    치아 영역 추출부는 제1 메쉬 데이터를 생성하기 이전에, 제1 볼륨 데이터로부터 추출된 치아 영역에 대하여 3차원 가우시안 필터를 적용하여 표면을 매끄럽게 하는, 치아 분할 시스템.The tooth region extracting unit applies a three-dimensional Gaussian filter to the tooth region extracted from the first volume data before generating the first mesh data to smooth the surface.
  6. 청구항 1 또는 청구항 2에 있어서,The method according to claim 1 or 2,
    인접 치아 분할부가 제1 메쉬 데이터로부터 개별 치아에 대한 제2 메쉬 데이터를 생성하는 것은, 제1 메쉬 데이터의 치아 영역에서 분할하고자 하는 각각의 개별 치아 위에 임의의 한 점을 입력받고, 이러한 점을 소스 또는 터미널로 지정하고, 그래프 분할 알고리즘을 적용하여 각 치아를 분할하는 측지선을 도출함으로써 수행되는, 치아 분할 시스템.When the adjacent tooth division unit generates the second mesh data for the individual teeth from the first mesh data, an arbitrary point is input on each individual tooth to be divided in the tooth area of the first mesh data, and this point is used as a source or by designating as a terminal, and applying a graph segmentation algorithm to derive geodesic lines segmenting each tooth, a tooth segmentation system.
  7. 청구항 2에 있어서, 3. The method according to claim 2,
    인접 치아 분할부가 제2 메쉬 데이터를 개별 치아에 대한 제2 볼륨 데이터로 변환하는 것은 제2 메쉬 데이터에 대하여 개별 치아에 대한 치아축, 치아 바운딩박스, 및 크기에 대한 수치정보를 도출함으로써 수행되는, 치아 분할 시스템.The adjacent tooth division unit converts the second mesh data into the second volume data for the individual teeth by deriving numerical information about the tooth axis, the tooth bounding box, and the size of the individual teeth with respect to the second mesh data. tooth division system.
  8. CT 촬영된 구강 이미지를 제1 볼륨 데이터로 입력받는 입력단계;An input step of receiving a CT-photographed oral image as first volume data;
    입력단계에서 입력된 제1 볼륨 데이터로부터 의미론적 분할(Semantic Segmentation) 방법을 사용하여 치아 영역을 추출하고, 추출된 치아 영역에 대하여 제1 메쉬 데이터를 생성하는 치아 영역 추출 단계;a tooth region extraction step of extracting a tooth region using a semantic segmentation method from the first volume data input in the input step, and generating first mesh data for the extracted tooth region;
    제1 메쉬 데이터로부터 개별 치아에 대한 제2 메쉬 데이터를 생성하여 개별 치아를 분할하는 인접 치아 분할 단계; 및an adjacent tooth segmentation step of dividing the individual teeth by generating second mesh data for the individual teeth from the first mesh data; and
    개별 치아에 대한 제2 메쉬 데이터를 출력하는 출력 단계;an output step of outputting second mesh data for individual teeth;
    를 포함하는, 치아 분할 방법.Including, tooth segmentation method.
  9. 청구항 8에 있어서,9. The method of claim 8,
    인접 치아 분할 단계는 제1 메쉬 데이터로부터 개별 치아에 대하여 생성된 제2 메쉬 데이터를 개별 치아에 대한 제2 볼륨 데이터로 변환하여 개별 치아를 분할하고,The adjacent tooth segmentation step divides the individual teeth by converting the second mesh data generated for the individual teeth from the first mesh data into the second volume data for the individual teeth,
    출력 단계는 인접 치아 분할 단계에서 변환된 제2 볼륨 데이터를 출력하는, 치아 분할 방법.The output step outputs the second volume data converted in the adjacent tooth partitioning step.
  10. 청구항 8 또는 청구항 9에 있어서,10. The method according to claim 8 or 9,
    치아 영역 추출 단계에서 사용되는 의미론적 분할(Semantic Segmentation)은 마르코프 랜덤필드(Markov Random Fields) 방법, FCN(Fully Convolution network) 방법, U-net 방법, 및 Deeplab V3+ 방법 중에서 선택되는 어느 하나인, 치아 분할 방법.The semantic segmentation used in the tooth region extraction step is any one selected from the Markov Random Fields method, the Fully Convolution network (FCN) method, the U-net method, and the Deeplab V3+ method. split method.
  11. 청구항 8 또는 청구항 9에 있어서,10. The method according to claim 8 or 9,
    치아 영역 추출 단계는 마칭 큐브(marching cubes) 알고리즘을 사용하여 제1 메쉬 데이터를 생성하는, 치아 분할 방법.wherein the tooth region extraction step generates first mesh data using a marching cubes algorithm.
  12. 청구항 8 또는 청구항 9에 있어서,10. The method according to claim 8 or 9,
    치아 영역 추출 단계는 제1 메쉬 데이터를 생성하기 이전에, 제1 볼륨 데이터로부터 추출된 치아 영역에 대하여 3차원 가우시안 필터를 적용하여 표면을 매끄럽게 하는, 치아 분할 방법.The tooth region extraction step applies a three-dimensional Gaussian filter to the tooth region extracted from the first volume data before generating the first mesh data to smooth the surface.
  13. 청구항 8 또는 청구항 9에 있어서,10. The method according to claim 8 or 9,
    인접 치아 분할 단계에서 제1 메쉬 데이터로부터 개별 치아에 대한 제2 메쉬 데이터를 생성하는 것은, 제1 메쉬 데이터의 치아 영역에서 분할하고자 하는 각각의 개별 치아 위에 임의의 한 점을 입력받고, 이러한 점을 소스 또는 터미널로 지정하고, 그래프 분할 알고리즘을 적용하여 각 치아를 분할하는 측지선을 도출함으로써 수행되는, 치아 분할 방법.In the step of dividing adjacent teeth, generating second mesh data for individual teeth from the first mesh data receives an arbitrary point on each individual tooth to be divided in the tooth area of the first mesh data, and A tooth segmentation method performed by specifying as a source or terminal, and applying a graph segmentation algorithm to derive geodesic lines segmenting each tooth.
  14. 청구항 9에 있어서, 10. The method of claim 9,
    인접 치아 분할 단계에서 제2 메쉬 데이터를 개별 치아에 대한 제2 볼륨 데이터로 변환하는 것은 제2 메쉬 데이터에 대하여 개별 치아에 대한 치아축, 치아 바운딩박스, 및 크기에 대한 수치정보를 도출함으로써 수행되는, 치아 분할 방법.Transforming the second mesh data into the second volume data for individual teeth in the adjacent tooth segmentation step is performed by deriving numerical information about the tooth axis, the tooth bounding box, and the size of the individual teeth with respect to the second mesh data. , how to split teeth.
PCT/KR2021/012952 2020-11-20 2021-09-23 System and method for segmenting teeth in ct image WO2022108082A1 (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
KR1020200156994A KR20220069655A (en) 2020-11-20 2020-11-20 Tooth segmentation system and method
KR10-2020-0156994 2020-11-20

Publications (1)

Publication Number Publication Date
WO2022108082A1 true WO2022108082A1 (en) 2022-05-27

Family

ID=81709104

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/KR2021/012952 WO2022108082A1 (en) 2020-11-20 2021-09-23 System and method for segmenting teeth in ct image

Country Status (2)

Country Link
KR (1) KR20220069655A (en)
WO (1) WO2022108082A1 (en)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR102496189B1 (en) * 2022-08-23 2023-02-08 주식회사 큐브세븐틴 Shell type temporary crown automatic manufacturing apparatus, method, and program
KR20240043443A (en) * 2022-09-27 2024-04-03 오스템임플란트 주식회사 Method and system for segmentation of a 3d oral model

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20200175678A1 (en) * 2018-11-28 2020-06-04 Orca Dental AI Ltd. Dental image segmentation and registration with machine learning
WO2020127398A1 (en) * 2018-12-17 2020-06-25 Promaton Holding B.V. Automated semantic segmentation of non-euclidean 3d data sets using deep learning
KR20200108822A (en) * 2017-10-02 2020-09-21 프로메이톤 홀딩 비.브이. Automatic classification and classification system of 3D tooth data using deep learning method
KR20200123160A (en) * 2018-02-16 2020-10-28 쓰리세이프 에이/에스 Intra-premises scanning with surface differentiation

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR101862378B1 (en) 2016-05-13 2018-05-29 연세대학교 산학협력단 Fully automatic dental segmentation system and method for computerized tomography

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20200108822A (en) * 2017-10-02 2020-09-21 프로메이톤 홀딩 비.브이. Automatic classification and classification system of 3D tooth data using deep learning method
KR20200123160A (en) * 2018-02-16 2020-10-28 쓰리세이프 에이/에스 Intra-premises scanning with surface differentiation
US20200175678A1 (en) * 2018-11-28 2020-06-04 Orca Dental AI Ltd. Dental image segmentation and registration with machine learning
WO2020127398A1 (en) * 2018-12-17 2020-06-25 Promaton Holding B.V. Automated semantic segmentation of non-euclidean 3d data sets using deep learning

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
EVANGELOS KALOGERAKIS; MELINOS AVERKIOU; SUBHRANSU MAJI; SIDDHARTHA CHAUDHURI: "3D Shape Segmentation with Projective Convolutional Networks", ARXIV.ORG, 8 December 2016 (2016-12-08), pages 1 - 11, XP080737954, DOI: 10.1109/CVPR.2017.702 *
YANG SHUAI, WANG RUIKUN, ZHAO WENJIE, KE YONGZHEN: "3D Intelligent Scissors for Dental Mesh Segmentation", COMPUTATIONAL AND MATHEMATICAL METHODS IN MEDICINE, vol. 2020, 31 January 2020 (2020-01-31), pages 1 - 12, XP055932217, ISSN: 1748-670X, DOI: 10.1155/2020/1394231 *

Also Published As

Publication number Publication date
KR20220069655A (en) 2022-05-27

Similar Documents

Publication Publication Date Title
WO2022164126A1 (en) Device and method for automatically matching oral scan data and computed tomography image by means of crown segmentation of oral scan data
WO2022108082A1 (en) System and method for segmenting teeth in ct image
CN110473192B (en) Digestive tract endoscope image recognition model training and recognition method, device and system
WO2016108453A1 (en) Teeth image automatic matching method, apparatus therefor, and recording medium therefor
CN110458830B (en) Image processing method, image processing apparatus, server, and storage medium
JPH09508817A (en) Method and apparatus for automated detection of gross abnormalities and asymmetries in chest images
CN112541893B (en) Method for detecting tree-structure bifurcation key points in three-dimensional tomography image
CN108765392B (en) Digestive tract endoscope lesion detection and identification method based on sliding window
JPS6362074A (en) Extracting device for connection component of three-dimensional image
WO2022131419A1 (en) Method for determining registration accuracy of three-dimensional dental ct image and three-dimensional digital impression model, and computer-readable recording medium in which program for executing same in computer is recorded
WO2022177095A1 (en) Artificial intelligence-based method and application for manufacturing 3d prosthesis for tooth restoration
CN114757960B (en) Tooth segmentation and reconstruction method based on CBCT image and storage medium
CN110210483A (en) Medical image lesion region dividing method, model training method and equipment
KR102461343B1 (en) Automatic tooth landmark detection method and system in medical images containing metal artifacts
WO2021025296A1 (en) Method for automatically recommending crown model and prosthesis cad apparatus for performing same
CN113344950A (en) CBCT image tooth segmentation method combining deep learning with point cloud semantics
WO2020004740A1 (en) Three-dimensional plane extraction method and device therefor
CN110136140A (en) Eye fundus image blood vessel image dividing method and equipment
WO2022010075A1 (en) Method for analyzing human tissue on basis of medical image and device thereof
WO2021006471A1 (en) Implant surgery planning method using automatic placement of implant structure, user interface providing method therefor, and teeth image processing device therefor
WO2023182702A1 (en) Artificial intelligence diagnosis data processing device and method for digital pathology images
WO2023008849A1 (en) Three-dimensional tooth image display apparatus and method, and method for training three-dimensional tooth image display apparatus
WO2014204126A2 (en) Apparatus for capturing 3d ultrasound images and method for operating same
CN114037744A (en) Oral cavity region-of-interest three-dimensional reconstruction method and device based on convolutional neural network
WO2020101265A1 (en) Myocardium image analysis method and device

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 21894855

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 21894855

Country of ref document: EP

Kind code of ref document: A1