WO2022182178A1 - Dispositif et procédé de diagnostic d'hémorragie cérébrale basé sur un apprentissage profond - Google Patents

Dispositif et procédé de diagnostic d'hémorragie cérébrale basé sur un apprentissage profond Download PDF

Info

Publication number
WO2022182178A1
WO2022182178A1 PCT/KR2022/002736 KR2022002736W WO2022182178A1 WO 2022182178 A1 WO2022182178 A1 WO 2022182178A1 KR 2022002736 W KR2022002736 W KR 2022002736W WO 2022182178 A1 WO2022182178 A1 WO 2022182178A1
Authority
WO
WIPO (PCT)
Prior art keywords
cerebral hemorrhage
deep learning
lesion
images
image
Prior art date
Application number
PCT/KR2022/002736
Other languages
English (en)
Korean (ko)
Inventor
김원태
김동민
이명재
강신욱
박종협
박기훈
Original Assignee
(주) 제이엘케이
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 (주) 제이엘케이 filed Critical (주) 제이엘케이
Priority claimed from KR1020220024246A external-priority patent/KR20220121217A/ko
Publication of WO2022182178A1 publication Critical patent/WO2022182178A1/fr

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
    • 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]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • 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
    • G16H30/20—ICT specially adapted for the handling or processing of medical images for handling medical images, e.g. DICOM, HL7 or PACS
    • 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
    • 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
    • G16H50/50—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for simulation or modelling of medical disorders
    • 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
    • G16H50/70—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for mining of medical data, e.g. analysing previous cases of other patients

Definitions

  • the present invention relates to an apparatus and method for diagnosing cerebral hemorrhage based on deep learning.
  • CNNs convolutional neural networks
  • An object of the present invention is to provide a method and apparatus for diagnosing cerebral hemorrhage capable of detecting a cerebral hemorrhage lesion from a CT image.
  • a deep learning-based cerebral hemorrhage diagnosis method includes: acquiring a plurality of computed tomography (CT) images by an image acquisition unit photographing a patient's brain; detecting, by a lesion detector, a brain hemorrhage lesion region from the plurality of two-dimensional CT images through a deep learning model; calculating, by a cerebral hemorrhage probability calculator, a cerebral hemorrhage probability of the patient based on the detected region; and outputting, by the cerebral hemorrhage probability calculator, the cerebral hemorrhage probability.
  • CT computed tomography
  • the detecting of the cerebral hemorrhage lesion region may include generating, by the lesion detector, a 3D image using the plurality of 2D CT images.
  • the 3D image may have a width, a height, and a depth of (224, 224, 80).
  • the method for diagnosing cerebral hemorrhage may further include storing the detection result of the cerebral hemorrhage lesion region as an image and storing the cerebral hemorrhage probability.
  • the deep learning model may be composed of a 3D-Unet using a convolution network and a deconvolution network.
  • An apparatus for diagnosing cerebral hemorrhage based on deep learning includes: an image acquisition unit configured to acquire a plurality of CT images by imaging a patient's brain; a lesion detector for detecting a brain hemorrhage lesion region from the plurality of two-dimensional CT images through a deep learning model; and a cerebral hemorrhage probability calculator for calculating and outputting the cerebral hemorrhage probability of the patient based on the detected region by the cerebral hemorrhage probability calculator.
  • the apparatus for diagnosing cerebral hemorrhage may further include a storage unit configured to store the detection result of the cerebral hemorrhage lesion region as an image and to store the cerebral hemorrhage probability.
  • cerebral hemorrhage can be accurately diagnosed only with CT images, which are cheaper than MRI images and can be taken quickly. Accordingly, in the case of an emergency patient whose time is short, a diagnosis of cerebral hemorrhage can be made quickly.
  • FIG. 1 is a block diagram of an apparatus for diagnosing cerebral hemorrhage according to an embodiment of the present invention.
  • FIG. 2 is a flowchart of a method for diagnosing cerebral hemorrhage and brain tumor lesion according to an embodiment of the present invention.
  • FIG. 3 is a diagram schematically illustrating a method of extracting a cerebral hemorrhagic lesion from a CT image according to an embodiment of the present invention.
  • FIG. 1 is a block diagram of an apparatus for diagnosing cerebral hemorrhage according to an embodiment of the present invention.
  • the cerebral hemorrhage diagnosis apparatus 100 includes an image acquisition unit 110 , a lesion detection unit 120 , a cerebral hemorrhage probability calculation unit 130 , a storage unit 160 , an output unit 170 , and a learning unit 180 . ) is included.
  • the image acquisition unit 110 may acquire computerized tomography (CT) images of the patient's brain.
  • CT images are tomographic images.
  • the lesion detector 120 may detect a brain hemorrhage lesion region from the plurality of 2D CT images through a deep learning model. To this end, the lesion detector 120 generates a single 3D image by summing all 2D CT images of the patient.
  • the 3D image has a width, a height, and a depth of (224, 224, 80). That is, the 3D image may be generated by summing 80 2D CT images.
  • the deep learning model may be composed of a 3D-Unet using a convolution network and a deconvolution network.
  • the cerebral hemorrhage probability calculator 130 may calculate a cerebral hemorrhage probability (probability) based on the detected cerebral hemorrhage lesion region. The cerebral hemorrhage probability calculator 130 may determine whether cerebral hemorrhage has occurred based on the calculated cerebral hemorrhage probability.
  • the storage unit 160 may store a software module for implementing the diagnostic method.
  • the storage unit 160 may store the extracted lesion information whenever the patient's image is updated.
  • the storage unit 160 may store the detected cerebral hemorrhage lesion region as an image. Also, the storage unit 160 may store the probability of cerebral hemorrhage.
  • the output unit 170 may display the image of the cerebral hemorrhage lesion region of each patient and the cerebral hemorrhage probability so that the user can check it.
  • the learning unit 180 may learn in advance the cerebral hemorrhage lesion from a plurality of CT images by deep learning.
  • the learning unit 180 has the same deep learning structure as the lesion extraction unit 120 , and may include a pooling structure for summing CNN and lesion information, a deconvolution structure for upsampling, and a skip connection structure for smooth learning. In other words, it is possible to utilize the deep learning structure in which learning and lesion extraction are the same.
  • the aforementioned image acquisition unit 110, lesion detection unit 120, cerebral hemorrhage probability calculation unit 130, storage unit 160, output unit 170, and learning unit 180 may be stored in a memory in the form of a software module.
  • the memory is connected to the processor, and the processor executes the software module to implement the functions of each software module independently or in a combined form to diagnose a lesion of at least one brain disease among cerebral hemorrhage and brain tumor, which will be described later. can be implemented.
  • the memory and the processor are components included in the computing device for implementing the method for diagnosing brain hemorrhage and brain tumor lesion.
  • the processor may be connected to various medical devices or databases via a network through a sub-communication system.
  • the cerebral hemorrhage diagnosis apparatus of the present embodiment learns about brain disease lesions including cerebral hemorrhage through artificial intelligence or machine learning, and provides automatically diagnosed cerebral disease lesions according to a preset operation process to a terminal of a doctor or hospital person. can do.
  • FIG. 2 is a flowchart of a method for diagnosing cerebral hemorrhage and brain tumor lesion according to an embodiment of the present invention.
  • the image acquisition unit 110 first acquires a plurality of two-dimensional CT images obtained by photographing the patient's brain ( S210 ). Thereafter, the pre-learned lesion extraction unit 120 extracts cerebral hemorrhage lesions using filters of different sizes ( S220 ). In this case, the lesion extraction unit 120 may extract the brain hemorrhage lesion through the deep learning model.
  • the deep learning model may be composed of a 3D-Unet using a convolution network and a deconvolution network.
  • the cerebral hemorrhage probability calculator 130 calculates the cerebral hemorrhage probability based on the detection result of the cerebral hemorrhage lesion region (S230).
  • the storage unit 160 may store the detected cerebral hemorrhage lesion region as an image and store the cerebral hemorrhage probability.
  • the cerebral hemorrhage probability calculator 130 may determine whether cerebral hemorrhage has occurred based on the calculated cerebral hemorrhage probability (S250).
  • FIG. 3 is a diagram schematically illustrating a method of extracting a cerebral hemorrhagic lesion from a CT image according to an embodiment of the present invention.
  • a cerebral hemorrhagic lesion is extracted from a plurality of CT images 310 obtained by photographing a patient's brain using a deep learning model.
  • the deep learning model is composed of a 3D-Unet using a convolution network and a deconvolution network.
  • Conv3D-BatchNormalization-Activation (selu) is basically used as one block.
  • Conv3D-BatchNormalization-Activation (selu) is used as one block in the deconvolutional network as in the convolutional network.
  • Conv3DTranspose In a deconvolutional network, a total of 3 Conv3DTranspose is applied.
  • the filter size of Conv3DTranspose is 3x3x1, and strides is 2x2x1.
  • Concatenate is used to connect the contents of the previous layer MaxPooling3D in the convolutional network.
  • each layer is connected by applying concatenate between the layers in the deconvolutional network a total of 3 times.
  • the filter size of Conv3D is 3x3x3 except for Conv3D just before the output of the model, and the filter size of the Conv3D output of the model is 1x1x1.
  • the deep learning model output is an image displayed by detecting a brain hemorrhagic lesion region from the input data, and has the same size as the input: 224 horizontal, 224 vertical, and 80 depth.

Landscapes

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

Abstract

L'invention concerne un procédé de diagnostic d'hémorragie cérébrale basé sur un apprentissage profond qui comprend : une étape dans laquelle une unité d'acquisition d'image acquiert une pluralité d'images de tomodensitométrie (TDM) par l'imagerie du cerveau d'un patient ; une étape dans laquelle une unité de détection de lésion détecte une région de lésion d'hémorragie cérébrale à partir de la pluralité d'images TDM bidimensionnelles par l'intermédiaire d'un modèle d'apprentissage profond ; une étape dans laquelle une unité de calcul de probabilité d'hémorragie cérébrale calcule la probabilité d'hémorragie cérébrale du patient sur la base de la région détectée ; et une étape dans laquelle l'unité de calcul de probabilité d'hémorragie cérébrale produit la probabilité d'hémorragie cérébrale.
PCT/KR2022/002736 2021-02-24 2022-02-24 Dispositif et procédé de diagnostic d'hémorragie cérébrale basé sur un apprentissage profond WO2022182178A1 (fr)

Applications Claiming Priority (4)

Application Number Priority Date Filing Date Title
KR10-2021-0024644 2021-02-24
KR20210024644 2021-02-24
KR10-2022-0024246 2022-02-24
KR1020220024246A KR20220121217A (ko) 2021-02-24 2022-02-24 딥러닝 기반의 뇌출혈 진단 장치 및 방법

Publications (1)

Publication Number Publication Date
WO2022182178A1 true WO2022182178A1 (fr) 2022-09-01

Family

ID=83049516

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/KR2022/002736 WO2022182178A1 (fr) 2021-02-24 2022-02-24 Dispositif et procédé de diagnostic d'hémorragie cérébrale basé sur un apprentissage profond

Country Status (1)

Country Link
WO (1) WO2022182178A1 (fr)

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20160082950A (ko) * 2016-06-27 2016-07-11 건국대학교 글로컬산학협력단 다중모드 ct를 이용하여 뇌졸중의 예후를 예측하는 방법
KR101740464B1 (ko) * 2016-10-20 2017-06-08 (주)제이엘케이인스펙션 뇌졸중 진단 및 예후 예측 방법 및 시스템
KR102015224B1 (ko) * 2018-09-18 2019-10-21 (주)제이엘케이인스펙션 딥러닝 기반의 뇌출혈 및 뇌종양 병변 진단 방법 및 장치
KR102165840B1 (ko) * 2020-05-21 2020-10-16 주식회사 휴런 인공지능 기반 뇌졸중 진단 장치 및 방법
KR20200135171A (ko) * 2019-05-24 2020-12-02 주식회사 루닛 의료 영상에서 악성의심 병변을 구별하는 방법, 이를 이용한 의료 영상 판독 방법 및 컴퓨팅 장치

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20160082950A (ko) * 2016-06-27 2016-07-11 건국대학교 글로컬산학협력단 다중모드 ct를 이용하여 뇌졸중의 예후를 예측하는 방법
KR101740464B1 (ko) * 2016-10-20 2017-06-08 (주)제이엘케이인스펙션 뇌졸중 진단 및 예후 예측 방법 및 시스템
KR102015224B1 (ko) * 2018-09-18 2019-10-21 (주)제이엘케이인스펙션 딥러닝 기반의 뇌출혈 및 뇌종양 병변 진단 방법 및 장치
KR20200135171A (ko) * 2019-05-24 2020-12-02 주식회사 루닛 의료 영상에서 악성의심 병변을 구별하는 방법, 이를 이용한 의료 영상 판독 방법 및 컴퓨팅 장치
KR102165840B1 (ko) * 2020-05-21 2020-10-16 주식회사 휴런 인공지능 기반 뇌졸중 진단 장치 및 방법

Similar Documents

Publication Publication Date Title
WO2018074739A1 (fr) Procédé et système de diagnostic et de pronostic d'un accident vasculaire cérébral
KR102015224B1 (ko) 딥러닝 기반의 뇌출혈 및 뇌종양 병변 진단 방법 및 장치
US8897533B2 (en) Medical image processing apparatus
WO2019231104A1 (fr) Procédé de classification d'images au moyen d'un réseau neuronal profond et appareil utilisant ledit procédé
Berkaya et al. Classification models for SPECT myocardial perfusion imaging
CN110858399B (zh) 用于提供虚拟断层扫描中风后续检查图像的方法和装置
WO2020138925A1 (fr) Procédé et système à base d'intelligence artificielle pour la classification d'une section de débit sanguin
US11076824B1 (en) Method and system for diagnosis of COVID-19 using artificial intelligence
WO2020122672A1 (fr) Appareil et procédé de segmentation automatique des vaisseaux sanguins par appariement d'une image fp et d'une image fag
WO2020138932A1 (fr) Procédé et système basés sur un apprentissage machine pour la classification de thrombus à l'aide d'une image d'écho gre
WO2019074172A1 (fr) Procédé et programme informatique pour analyser un indicateur quantitatif basé sur le volume pour la mesure de l'amyloïde dans une image cérébrale en tep
WO2019143179A1 (fr) Procédé de détection automatique de mêmes régions d'intérêt entre des images du même objet prises à un intervalle de temps, et appareil ayant recours à ce procédé
WO2015147595A1 (fr) Procédé pour distinguer un artéfact et une région malade dans des images médicales
US20230077125A1 (en) Method for diagnosing age-related macular degeneration and defining location of choroidal neovascularization
KR102245219B1 (ko) 의료 영상에서 악성의심 병변을 구별하는 방법, 이를 이용한 의료 영상 판독 방법 및 컴퓨팅 장치
Wang et al. Reduction of bubble-like frames using a RSS filter in wireless capsule endoscopy video
CN114445406B (zh) 肠镜图像分析方法、装置和医学图像处理设备
WO2020111534A2 (fr) Procédé et système pour estimer l'âge vasculaire cérébral à partir d'une image médicale
Lucassen et al. Deep learning for detection and localization of B-lines in lung ultrasound
WO2020179950A1 (fr) Procédé et dispositif à base d'apprentissage profond pour la prédiction de la progression d'une maladie cérébrale
US20230113721A1 (en) Functional measurements of vessels using a temporal feature
WO2022182178A1 (fr) Dispositif et procédé de diagnostic d'hémorragie cérébrale basé sur un apprentissage profond
CN114742836B (zh) 医学图像的处理方法、装置及计算机设备
CN110197722B (zh) Ai-cpu系统平台
CN111753831A (zh) 一种图像分析方法、装置、图像采集设备及存储介质

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: 22760094

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: 22760094

Country of ref document: EP

Kind code of ref document: A1