WO2020215807A1 - Procédé basé sur l'apprentissage profond pour améliorer le taux de détection de polype adénomateux par coloscope - Google Patents

Procédé basé sur l'apprentissage profond pour améliorer le taux de détection de polype adénomateux par coloscope Download PDF

Info

Publication number
WO2020215807A1
WO2020215807A1 PCT/CN2020/000063 CN2020000063W WO2020215807A1 WO 2020215807 A1 WO2020215807 A1 WO 2020215807A1 CN 2020000063 W CN2020000063 W CN 2020000063W WO 2020215807 A1 WO2020215807 A1 WO 2020215807A1
Authority
WO
WIPO (PCT)
Prior art keywords
polyp
polyps
colonoscopy
detection
improving
Prior art date
Application number
PCT/CN2020/000063
Other languages
English (en)
Chinese (zh)
Inventor
王玉峰
Original Assignee
天津御锦人工智能医疗科技有限公司
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 天津御锦人工智能医疗科技有限公司 filed Critical 天津御锦人工智能医疗科技有限公司
Publication of WO2020215807A1 publication Critical patent/WO2020215807A1/fr

Links

Images

Classifications

    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B1/00—Instruments for performing medical examinations of the interior of cavities or tubes of the body by visual or photographical inspection, e.g. endoscopes; Illuminating arrangements therefor
    • A61B1/31—Instruments for performing medical examinations of the interior of cavities or tubes of the body by visual or photographical inspection, e.g. endoscopes; Illuminating arrangements therefor for the rectum, e.g. proctoscopes, sigmoidoscopes, colonoscopes
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B1/00—Instruments for performing medical examinations of the interior of cavities or tubes of the body by visual or photographical inspection, e.g. endoscopes; Illuminating arrangements therefor
    • A61B1/00002—Operational features of endoscopes
    • A61B1/00004—Operational features of endoscopes characterised by electronic signal processing
    • A61B1/00009—Operational features of endoscopes characterised by electronic signal processing of image signals during a use of endoscope
    • A61B1/000094—Operational features of endoscopes characterised by electronic signal processing of image signals during a use of endoscope extracting biological structures
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B1/00—Instruments for performing medical examinations of the interior of cavities or tubes of the body by visual or photographical inspection, e.g. endoscopes; Illuminating arrangements therefor
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B1/00—Instruments for performing medical examinations of the interior of cavities or tubes of the body by visual or photographical inspection, e.g. endoscopes; Illuminating arrangements therefor
    • A61B1/00002—Operational features of endoscopes
    • A61B1/00004—Operational features of endoscopes characterised by electronic signal processing
    • A61B1/00009—Operational features of endoscopes characterised by electronic signal processing of image signals during a use of endoscope
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B1/00—Instruments for performing medical examinations of the interior of cavities or tubes of the body by visual or photographical inspection, e.g. endoscopes; Illuminating arrangements therefor
    • A61B1/00002—Operational features of endoscopes
    • A61B1/00004—Operational features of endoscopes characterised by electronic signal processing
    • A61B1/00009—Operational features of endoscopes characterised by electronic signal processing of image signals during a use of endoscope
    • A61B1/000096—Operational features of endoscopes characterised by electronic signal processing of image signals during a use of endoscope using artificial intelligence
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B1/00—Instruments for performing medical examinations of the interior of cavities or tubes of the body by visual or photographical inspection, e.g. endoscopes; Illuminating arrangements therefor
    • A61B1/00002—Operational features of endoscopes
    • A61B1/00043—Operational features of endoscopes provided with output arrangements
    • A61B1/00045—Display arrangement
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B1/00—Instruments for performing medical examinations of the interior of cavities or tubes of the body by visual or photographical inspection, e.g. endoscopes; Illuminating arrangements therefor
    • A61B1/00002—Operational features of endoscopes
    • A61B1/00043—Operational features of endoscopes provided with output arrangements
    • A61B1/00055—Operational features of endoscopes provided with output arrangements for alerting the user
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B1/00—Instruments for performing medical examinations of the interior of cavities or tubes of the body by visual or photographical inspection, e.g. endoscopes; Illuminating arrangements therefor
    • A61B1/04—Instruments for performing medical examinations of the interior of cavities or tubes of the body by visual or photographical inspection, e.g. endoscopes; Illuminating arrangements therefor combined with photographic or television appliances
    • 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
    • G16H30/40—ICT specially adapted for the handling or processing of medical images for processing medical images, e.g. editing

Definitions

  • the invention relates to the technical field of colonoscopy adenomatous polyps detection, in particular to a method for improving the detection rate of colonoscopy adenomatous polyps based on deep learning.
  • Polyps are diseased tissues that protrude or bulge on the mucosal surface of the intestinal cavity. With the help of colonoscopy, the size and number of polyps can be detected. According to the pathological classification, polyps are divided into inflammatory polyps, hyperplastic polyps, hamartomas, and adenomatous polyps. Among them, adenomatous polyps are more common, accounting for about 70%-80%, and the size is generally about 0.5-2cm. In addition to the pathological classification of adenomatous polyp carcinogenesis, it is generally believed that the size and number of adenomas have a great influence on the possibility of canceration.
  • the carcinogenesis rate of adenomatous polyps less than 1cm is almost zero, and the chance of adenomatous polyps greater than 1.0cm is increased.
  • the carcinogenesis rate of 1-2cm adenomatous polyps is about 10%, and that of adenomatous polyps >2m
  • the rate is as high as 50%. According to statistics, the number of polyps is less than 3, and the canceration rate is 12%-29%; for 3 or more polyps, the canceration rate increases to 66.7%.
  • Colonoscopy This is the most sensitive test for detecting colon polyps and colon cancer. It is similar to a sigmoidoscopy, but the instrument used (i.e. colonoscope) is a longer and thin tube connected to the camera and control part, so the doctor can use it to examine your rectum and the entire colon. If any polyp is found during the examination, the doctor can remove it immediately or take a part of the tissue for biopsy.
  • Fecal occult blood test This non-invasive test is used to detect whether your stool contains blood.
  • Capsule endoscopy The medical community has now invented a capsule with a miniature camera inside. After swallowing it, it can distinguish polyps in the small intestine with high accuracy. However, because small intestinal polyps are relatively rare, this test is not commonly used.
  • Colonoscopy is also the most widely used and most effective polyp screening method.
  • Colonoscopy only relying on the naked eye to find some polyps during the operation will often lead to a missed detection. Therefore, increasing the detection rate of adenomatous polyps in colonoscopy has become a top priority .
  • the purpose of the present invention is to provide a method for improving the detection rate of colonoscopy adenomatous polyps based on deep learning in view of the technical defects in the prior art, which is used to solve the problem of relying on manual detection of polyps in the process of traditional colonoscopy. , It is easy to miss the detection problem due to the negligence of the doctor or the small size of the polyp.
  • a method for improving the detection rate of colonoscopy adenomatous polyps based on deep learning includes the following steps:
  • the polyp detection model detects whether there are polyps and the probability of polyps in each frame of image
  • the polyp detection model is obtained through the following steps:
  • the training set is used to train the formed initialization model
  • the test set is used to test
  • the polyp detection model is finally obtained through training and testing.
  • the polyp detection model is constructed using the YOLOv3 detection algorithm.
  • This invention uses artificial intelligence deep neural network, combined with medical big data and medical knowledge, can automatically detect polyps that appear in the lens during colonoscopy surgery, and improve the recognition rate of polyps during colonoscopy, thereby indirectly improving The detection rate of adenomatous polyps.
  • Figure 1 is a flow chart of video stream transmission during colonoscopy
  • Figure 2 is a training flowchart of a polyp detection model.
  • a method for improving the detection rate of colonoscopy adenomatous polyps based on deep learning of the present invention includes the following steps:
  • Step 1 When the operation starts, the video stream from the enteroscopy lens in the operating table is divided into two, one part is transmitted to the doctor's operating platform, and the other part is sent to the polyp detection model formed based on convolutional neural network training (embedded Identify in the artificial intelligence detection module of the colonoscopy operating system);
  • Step 2 Preprocess the video stream, and then send it to the polyp detection model for identification. For each frame of the video stream, detect whether there is a polyp and the probability that the detected target is a polyp.
  • Step 3 Return the test result of step 2 to the doctor's operating platform for display.
  • Step 4 If a polyp appears in the video stream, frame it to prompt.
  • the overall flow chart of the present invention is shown in Figure 1, where the polyp detection model formed based on convolutional neural network training in step 2 is developed using the YOLOv3 detection algorithm to form a polyp detection model to meet the requirements of the entire colonoscopy. The real-time required.
  • the training of the polyp detection model formed based on the training of the convolutional neural network specifically includes the following steps;
  • Step 1 Obtain a clear and bright image set with polyps that was intercepted during colonoscopy from the hospital database;
  • Step 2 Label the image set obtained in step 1, and label the target detection objects such as polyps in the image using labelimg.
  • the labeled image set is divided into two parts: training set and test set, which is convenient for training, including:
  • Step 2.1 uniformly crop the image set obtained in step 1, and crop it into an image set of uniform size and same format;
  • Step 2.2 Use labelimg to label the target in the image set to obtain a complete polyp image set
  • Step 2.3 Select 1500 images in the complete image set as the training set and 300 images as the test set;
  • Step 3 Input the selected training set into the initialization model of YOLOv3, set the parameters in the training process, and then train;
  • Step 4 Save the model obtained after training
  • Step 5 Transmit the test set as input to the trained model, and detect the output result of the learning network
  • Step 6 Embed the trained model into the complete colonoscopy operating system for real-time detection during the operation.
  • the invention reduces the missed detection rate of polyps in the traditional traditional colonoscopy operation by using the higher accuracy of deep learning, and provides more reliable and efficient support for the diagnosis of doctors.
  • the present invention uses deep learning technology to transmit the video output from the colonoscopy lens to the trained neural network, automatically detects some polyps that appear inside the lens business during the colonoscopy operation, and reminds the doctor to perform further operations, thereby improving The detection rate of adenomatous polyps.

Landscapes

  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Surgery (AREA)
  • Engineering & Computer Science (AREA)
  • Medical Informatics (AREA)
  • Public Health (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • General Health & Medical Sciences (AREA)
  • Physics & Mathematics (AREA)
  • Radiology & Medical Imaging (AREA)
  • Pathology (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Biomedical Technology (AREA)
  • Molecular Biology (AREA)
  • Animal Behavior & Ethology (AREA)
  • Optics & Photonics (AREA)
  • Biophysics (AREA)
  • Veterinary Medicine (AREA)
  • Signal Processing (AREA)
  • Evolutionary Computation (AREA)
  • Artificial Intelligence (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Epidemiology (AREA)
  • Primary Health Care (AREA)
  • Image Analysis (AREA)

Abstract

L'invention concerne un procédé basé sur un apprentissage profond pour améliorer un taux de détection de polype adénomateux par coloscope. Le procédé comprend les étapes suivantes consistant à : diviser un flux vidéo transmis par une caméra d'un coloscope d'une table d'opération en deux parties, une partie du flux vidéo étant transmise à une plate-forme opérationnelle d'un médecin et l'autre partie du flux vidéo étant prétraitée puis transmise à un modèle de détection de polype inclus dans un système d'exploitation de coloscope pour l'identification ; le modèle de détection de polype détecte s'il y a un polype dans chaque trame d'image et la probabilité d'occurrence du polype ; renvoyer un résultat de détection du modèle de détection de polype à la plate-forme opérationnelle du médecin et l'afficher ; et s'il y a un polype dans le flux vidéo, cadrer le polype et présenter une notification. Grâce à un réseau neuronal profond d'intelligence artificielle, un polype se trouvant dans le rayon d'action d'une caméra d'un coloscope pendant une opération peut être automatiquement détecté, ce qui permet d'améliorer le taux d'identification d'un polype pendant une coloscopie et d'améliorer ainsi indirectement le taux de détection d'un polype adénomateux.
PCT/CN2020/000063 2019-04-25 2020-04-09 Procédé basé sur l'apprentissage profond pour améliorer le taux de détection de polype adénomateux par coloscope WO2020215807A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910340477.7A CN111839428A (zh) 2019-04-25 2019-04-25 一种基于深度学习提高结肠镜腺瘤性息肉检出率的方法
CN201910340477.7 2019-04-25

Publications (1)

Publication Number Publication Date
WO2020215807A1 true WO2020215807A1 (fr) 2020-10-29

Family

ID=72940927

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/000063 WO2020215807A1 (fr) 2019-04-25 2020-04-09 Procédé basé sur l'apprentissage profond pour améliorer le taux de détection de polype adénomateux par coloscope

Country Status (2)

Country Link
CN (1) CN111839428A (fr)
WO (1) WO2020215807A1 (fr)

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112785549B (zh) * 2020-12-29 2024-03-01 成都微识医疗设备有限公司 基于图像识别的肠镜检查质量评估方法、装置及存储介质
CN112669283B (zh) * 2020-12-29 2022-11-01 杭州优视泰信息技术有限公司 一种基于深度学习的肠镜图像息肉误检测抑制装置
CN112598086A (zh) * 2021-03-04 2021-04-02 四川大学 基于深度神经网络的常见结肠部疾病分类方法及辅助系统
CN113284146B (zh) * 2021-07-23 2021-10-22 天津御锦人工智能医疗科技有限公司 结直肠息肉图像的识别方法、装置及存储介质

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107256552A (zh) * 2017-06-14 2017-10-17 成都康托医疗设备有限公司 息肉图像识别系统及方法
CN108292366A (zh) * 2015-09-10 2018-07-17 美基蒂克艾尔有限公司 在内窥镜手术中检测可疑组织区域的系统和方法
CN109523535A (zh) * 2018-11-15 2019-03-26 首都医科大学附属北京友谊医院 一种病变图像的预处理方法
US20190095760A1 (en) * 2017-09-27 2019-03-28 Fujifilm Corporation Learning assistance device, method of operating learning assistance device, learning assistance program, learning assistance system, and terminal device
CN109635866A (zh) * 2018-12-10 2019-04-16 杭州帝视科技有限公司 处理肠图像的方法
CN109671053A (zh) * 2018-11-15 2019-04-23 首都医科大学附属北京友谊医院 一种胃癌图像识别系统、装置及其应用

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109447973B (zh) * 2018-10-31 2021-11-26 腾讯医疗健康(深圳)有限公司 一种结肠息肉图像的处理方法和装置及系统

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108292366A (zh) * 2015-09-10 2018-07-17 美基蒂克艾尔有限公司 在内窥镜手术中检测可疑组织区域的系统和方法
CN107256552A (zh) * 2017-06-14 2017-10-17 成都康托医疗设备有限公司 息肉图像识别系统及方法
US20190095760A1 (en) * 2017-09-27 2019-03-28 Fujifilm Corporation Learning assistance device, method of operating learning assistance device, learning assistance program, learning assistance system, and terminal device
CN109523535A (zh) * 2018-11-15 2019-03-26 首都医科大学附属北京友谊医院 一种病变图像的预处理方法
CN109671053A (zh) * 2018-11-15 2019-04-23 首都医科大学附属北京友谊医院 一种胃癌图像识别系统、装置及其应用
CN109635866A (zh) * 2018-12-10 2019-04-16 杭州帝视科技有限公司 处理肠图像的方法

Also Published As

Publication number Publication date
CN111839428A (zh) 2020-10-30

Similar Documents

Publication Publication Date Title
WO2020215807A1 (fr) Procédé basé sur l'apprentissage profond pour améliorer le taux de détection de polype adénomateux par coloscope
CN109146884B (zh) 内窥镜检查监控方法及装置
WO2021036863A1 (fr) Système d'aide au diagnostic basé sur l'apprentissage profond pour un cancer précoce du tractus digestif et appareil d'examen
WO2019088121A1 (fr) Appareil d'aide au diagnostic d'image, procédé de collecte de données, procédé d'aide au diagnostic d'image et programme d'assistance au diagnostic d'image
CN107256552B (zh) 息肉图像识别系统及方法
AU2019431299B2 (en) AI systems for detecting and sizing lesions
US8423123B2 (en) System and method for in-vivo feature detection
US7885446B2 (en) System and method to detect a transition in an image stream
CN113164010A (zh) 利用消化器官的内窥镜影像的疾病的诊断支持方法、诊断支持系统、诊断支持程序及存储了该诊断支持程序的计算机可读记录介质
WO2020215810A1 (fr) Procédé de détection d'image à bande étroite basé sur la reconnaissance d'image pour procédure de coloscopie
CN108615037A (zh) 基于深度学习的可控胶囊内镜操作实时辅助系统及操作方法
WO2023143014A1 (fr) Procédé et dispositif d'inspection assistée par endoscope basés sur l'intelligence artificielle
CN112785549B (zh) 基于图像识别的肠镜检查质量评估方法、装置及存储介质
WO2021139672A1 (fr) Procédé, appareil et dispositif d'assistance d'opération médicale, et support de stockage informatique
Li et al. Intelligent detection endoscopic assistant: An artificial intelligence-based system for monitoring blind spots during esophagogastroduodenoscopy in real-time
CN114446435A (zh) 一种肠镜报告生成方法和装置
Song et al. Use of artificial intelligence to improve the quality control of gastrointestinal endoscopy
WO2024164912A1 (fr) Système et procédé d'évaluation de structure cible endoscopique, dispositif et support de stockage
CN116309605B (zh) 基于深度学习和状态转移的内窥镜检查质控方法及系统
CN114359131A (zh) 幽门螺杆菌胃部视频全自动智能分析系统及其标记方法
CN108937871A (zh) 一种消化道微光学相干断层成像图像分析系统及方法
Gan et al. A feasibility trial of computer-aided diagnosis for enteric lesions in capsule endoscopy
CN111839429A (zh) 一种基于图像识别判断结肠镜检查是否完整的检测方法
CN111863177A (zh) 一种基于大数据的肠镜检查质量报告管理系统
CN110772210A (zh) 一种诊断交互系统及方法

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: 20794572

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: 20794572

Country of ref document: EP

Kind code of ref document: A1