Endoscopy 2023; 55(S 02): S8-S9
DOI: 10.1055/s-0043-1765009
Abstracts | ESGE Days 2023
Oral presentation
AI: an expert endoscopist's eye in every room? 20/04/2023, 08:30 – 09:30 Liffey Meeting Room 2

Computer-aided diagnosis (CADx) improves characterization of Barrett’s neoplasia by endoscopists

Authors

  • J. Jelmer

    1   Amsterdam UMC, locatie VUmc, Amsterdam, Netherlands

  • K. Kusters

    2   Eindhoven University of Technology, Eindhoven, Netherlands

  • M. Jong

    1   Amsterdam UMC, locatie VUmc, Amsterdam, Netherlands

  • K. Fockens

    1   Amsterdam UMC, locatie VUmc, Amsterdam, Netherlands

  • T. Boers

    2   Eindhoven University of Technology, Eindhoven, Netherlands

  • J. Van Der Putten

    2   Eindhoven University of Technology, Eindhoven, Netherlands

  • M. Struyvenberg

    1   Amsterdam UMC, locatie VUmc, Amsterdam, Netherlands

  • L. Duits

    1   Amsterdam UMC, locatie VUmc, Amsterdam, Netherlands

  • B.L.A. M. Weusten

    3   St. Antonius Hospital, Nieuwegein, Netherlands
    4   UMC Utrecht, Utrecht, Netherlands

  • L. Alvarez Herrero

    3   St. Antonius Hospital, Nieuwegein, Netherlands

  • M.H.M. G. Houben

    5   Haga Ziekenhuis, Den Haag, Netherlands

  • W. B. Nagengast

    6   University Medical Center Groningen, Groningen, Netherlands

  • J. Westerhof

    6   University Medical Center Groningen, Groningen, Netherlands

  • A. Alkhalaf

    7   Isala Zwolle, Zwolle, Netherlands

  • R. Mallant

    8   FlevoHospital Almere, Almere, Netherlands

  • P. Scholten

    9   OLVG, location West, Amsterdam, Netherlands

  • K. Ragunath

    10   Royal Perth Hospital, Perth, Australia

  • S. Seewald

    11   Hirslanden Klinik Hirslanden, Zürich, Switzerland

  • P. Elbe

    12   Karolinska University Hospital, Stockholm, Sweden
    13   Karolinska Institutet Clintec, Stockholm, Sweden

  • F. Baldaque-Silva

    12   Karolinska University Hospital, Stockholm, Sweden

  • M. Barret

    14   Cochin Hospital, Paris, France

  • J.Ortiz Fernández-Sordo

    15   Queen's Medical Centre, Nottingham, United Kingdom

  • G. Moral Villarejo

    15   Queen's Medical Centre, Nottingham, United Kingdom

  • O. Pech

    16   Krankenhaus Barmherzige Brüder Regensburg, Regensburg, Germany

  • T. Beyna

    17   Evangelisches Krankenhaus Düsseldorf, Kirchfeldstraße, Düsseldorf, Germany

  • F. Van Der Sommen

    2   Eindhoven University of Technology, Eindhoven, Netherlands

  • P. De With

    2   Eindhoven University of Technology, Eindhoven, Netherlands

  • J. De Groof

    1   Amsterdam UMC, locatie VUmc, Amsterdam, Netherlands

  • J.J.G.H. M. Bergman

    1   Amsterdam UMC, locatie VUmc, Amsterdam, Netherlands

  • R. E. Pouw

    1   Amsterdam UMC, locatie VUmc, Amsterdam, Netherlands
Further Information
Also available at
 
 

    Aims Characterization of visible abnormalities in Barrett esophagus (BE) patients can be challenging, resulting in suboptimal diagnostic accuracy and poor inter-observer agreement. Computer-aided diagnosis (CADx) systems may assist endoscopists in neoplasia characterization. We aimed to develop and validate a CADx system for BE neoplasia.

    Methods The CADx system received pretraining with ImageNet with consecutive domain-specific pretraining with GastroNet which includes >5 million endoscopy images. This pre-trained system was subsequently trained and internally validated using 1754 narrow-band imaging (NBI) images of early BE neoplasia (348 patients) and 1838 NBI images of non-dysplastic BE (197 patients), obtained in 8 international centers. The CADx system was tested on a test set which consisted of 30 videos (20 patients) of BE neoplasia and 60 videos (31 patients) of non-dysplastic BE. The test set was benchmarked by 44 general endoscopists without and with CADx assistance. Finally 10 international independent BE experts evaluated the test set.

    Results The performances of CADx and endoscopists are presented in (▶Table [1]). The stand-alone sensitivity of the CADx system was superior to general endoscopists (p = 0.04). Assistance of the CADx system significantly increased the sensitivity (p<0.001) and specificity (p<0.001) of general endoscopists to the level of expert endoscopists. CADx assistance furthermore increased endoscopists’ confidence in characterization (P<0.001).

    Zoom
    Table 1  

    Conclusions CADx assistance significantly increased characterization performance of BE neoplasia by general endoscopists to the level of expert endoscopists. The use of this CADx system may thereby improve daily Barrett surveillance.

    Table 1: Performances of CADx and endoscopists on test sets


     

    Conflicts of interest

    This project was funded by Olympus Endoscopy (Tokyo, Japan)

    Publication History

    Article published online:
    14 April 2023

    © 2023. European Society of Gastrointestinal Endoscopy. All rights reserved.

    Georg Thieme Verlag KG
    Rüdigerstraße 14, 70469 Stuttgart, Germany

     
    Zoom
    Table 1