Back
Full-text
Journal article

An objective comparison of detection and segmentation algorithms for artefacts in clinical endoscopy.

  • Ali S Institute of Biomedical Engineering, Big Data Institute, Department of Engineering Science, University of Oxford, Oxford, UK. sharib.ali@eng.ox.ac.uk.
  • Zhou F Ludwig Institute for Cancer Research, University of Oxford, Oxford, UK.
  • Braden B Translational Gastroenterology Unit, Nuffield Department of Medicine, Experimental Medicine Div., John Radcliffe Hospital, University of Oxford, Oxford, UK.
  • Bailey A Translational Gastroenterology Unit, Nuffield Department of Medicine, Experimental Medicine Div., John Radcliffe Hospital, University of Oxford, Oxford, UK.
  • Yang S Ping An Technology (Shenzhen) Co. Ltd, Shenzhen, China.
  • Cheng G Ping An Technology (Shenzhen) Co. Ltd, Shenzhen, China.
  • Zhang P Beijing Institute of Technology, Beijing, China.
  • Li X Beijing Institute of Technology, Beijing, China.
  • Kayser M Technishe Universität München, Munich, Germany.
  • Soberanis-Mukul RD Technishe Universität München, Munich, Germany.
  • Albarqouni S Technishe Universität München, Munich, Germany.
  • Wang X Department of Biomedical Engineering, University of California, Davis, USA.
  • Wang C Department of Ultrasound Imaging, Tiantan Hospital, Beijing, China.
  • Watanabe S Department of Bioinformatic Engineering, Osaka University, Suita, Osaka, Japan.
  • Oksuz I School of Biomedical Engineering and Imaging Sciences, King's College London, London, UK.
  • Ning Q Department of Automation, Shanghai Jiao Tong University, Shanghai, China.
  • Yang S School of Engineering, University of Glasgow, Glasgow, UK.
  • Khan MA Department of Computer Science and Engineering, Korea University, Seoul, South Korea.
  • Gao XW Department of Computer Science, Middlesex University, London, UK.
  • Realdon S Instituto Onclologico Veneto, IOV-IRCCS, Padova, Italy.
  • Loshchenov M A.M. Prokhorov General Physics Institute, Russian Academy of Science, Moscow, Russia.
  • Schnabel JA School of Biomedical Engineering and Imaging Sciences, King's College London, London, UK.
  • East JE Translational Gastroenterology Unit, Nuffield Department of Medicine, Experimental Medicine Div., John Radcliffe Hospital, University of Oxford, Oxford, UK.
  • Wagnieres G Swiss Federal Institute of Technology in Lausanne (EPFL), Lausanne, Switzerland.
  • Loschenov VB A.M. Prokhorov General Physics Institute, Russian Academy of Science, Moscow, Russia.
  • Grisan E Department of Information Engineering, University of Padova, Padova, Italy.
  • Daul C CRAN UMR 7039, University of Lorraine, CNRS, Nancy, France.
  • Blondel W CRAN UMR 7039, University of Lorraine, CNRS, Nancy, France.
  • Rittscher J Institute of Biomedical Engineering, Big Data Institute, Department of Engineering Science, University of Oxford, Oxford, UK.
Show more…
  • 2020-02-19
Published in:
  • Scientific reports. - 2020
Algorithms
Artifacts
Colon
Datasets as Topic
Endoscopy
Esophagus
Female
Humans
Image Interpretation, Computer-Assisted
Imaging, Three-Dimensional
International Cooperation
Male
Neural Networks, Computer
Stomach
Urinary Bladder
Uterus
English We present a comprehensive analysis of the submissions to the first edition of the Endoscopy Artefact Detection challenge (EAD). Using crowd-sourcing, this initiative is a step towards understanding the limitations of existing state-of-the-art computer vision methods applied to endoscopy and promoting the development of new approaches suitable for clinical translation. Endoscopy is a routine imaging technique for the detection, diagnosis and treatment of diseases in hollow-organs; the esophagus, stomach, colon, uterus and the bladder. However the nature of these organs prevent imaged tissues to be free of imaging artefacts such as bubbles, pixel saturation, organ specularity and debris, all of which pose substantial challenges for any quantitative analysis. Consequently, the potential for improved clinical outcomes through quantitative assessment of abnormal mucosal surface observed in endoscopy videos is presently not realized accurately. The EAD challenge promotes awareness of and addresses this key bottleneck problem by investigating methods that can accurately classify, localize and segment artefacts in endoscopy frames as critical prerequisite tasks. Using a diverse curated multi-institutional, multi-modality, multi-organ dataset of video frames, the accuracy and performance of 23 algorithms were objectively ranked for artefact detection and segmentation. The ability of methods to generalize to unseen datasets was also evaluated. The best performing methods (top 15%) propose deep learning strategies to reconcile variabilities in artefact appearance with respect to size, modality, occurrence and organ type. However, no single method outperformed across all tasks. Detailed analyses reveal the shortcomings of current training strategies and highlight the need for developing new optimal metrics to accurately quantify the clinical applicability of methods.
Language
  • English
Open access status
gold
Identifiers
Persistent URL
https://folia.unifr.ch/global/documents/16552
Statistics
Document views: 74 File downloads:
  • Full-text: 0