Video-based computer aided detection system improves Barrett’s neoplasia detection of general endoscopists in a multi-step benchmarking study

M. Jong; K. Fockens; J. Jukema; J. Van Der Putten; T. Boers; K. Kusters; R. E. Pouw; L. Duits; B.L.A. M. Weusten; L. Alvarez Herrero; M.H.M. G. Houben; W. B. Nagengast; J. Westerhof; A. Alkhalaf; R. Mallant; P. Scholten; K. Ragunath; S. Seewald; P. Elbe; F. Baldaque-Silva; M. Barret; J. Ortiz Fernández-Sordo; G. Moral Villarejo; O. Pech; T. Beyna; F. Van Der Sommen; P. De With; J. De Groof; J.J.G.H. M. Bergman

doi:10.1055/s-0043-1765005

Endoscopy, Table of Contents

Endoscopy 2023; 55(S 02): S7
DOI: 10.1055/s-0043-1765005

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

Video-based computer aided detection system improves Barrett’s neoplasia detection of general endoscopists in a multi-step benchmarking study

Authors

M. Jong

¹Amsterdam UMC, locatie VUmc, Amsterdam, Netherlands
K. Fockens

¹Amsterdam UMC, locatie VUmc, Amsterdam, Netherlands
J. Jukema

²Eindhoven University of Technology, Eindhoven, Netherlands
J. Van Der Putten

²Eindhoven University of Technology, Eindhoven, Netherlands
T. Boers

²Eindhoven University of Technology, Eindhoven, Netherlands
K. Kusters

²Eindhoven University of Technology, Eindhoven, Netherlands
R. E. Pouw

¹Amsterdam UMC, locatie VUmc, Amsterdam, Netherlands
L. Duits

¹Amsterdam UMC, locatie VUmc, Amsterdam, Netherlands
B.L.A. M. Weusten

³St. Antonius Hospital, Nieuwegein, Netherlands

⁴UMC Utrecht, Utrecht, Netherlands
L. Alvarez Herrero

³St. Antonius Hospital, Nieuwegein, Netherlands
M.H.M. G. Houben

⁵Haga Ziekenhuis, Den Haag, Netherlands
W. B. Nagengast

⁶University Medical Center Groningen, Groningen, Netherlands
J. Westerhof

⁶University Medical Center Groningen, Groningen, Netherlands
A. Alkhalaf

⁷Isala Zwolle, Zwolle, Netherlands
R. Mallant

⁸FlevoHospital Almere, Almere, Netherlands
P. Scholten

⁹OLVG, location West, Amsterdam, Netherlands
K. Ragunath

¹⁰Royal Perth Hospital, Perth, Australia
S. Seewald

¹¹Hirslanden Klinik Hirslanden, Zürich, Switzerland
P. Elbe

¹²Karolinska University Hospital, Stockholm, Sweden

¹³Karolinska Institutet Clintec, Stockholm, Sweden
F. Baldaque-Silva

¹²Karolinska University Hospital, Stockholm, Sweden
M. Barret

¹⁴Cochin Hospital, Paris, France
J. Ortiz Fernández-Sordo

¹⁵Queen's Medical Centre, Nottingham, United Kingdom
G. Moral Villarejo

¹⁵Queen's Medical Centre, Nottingham, United Kingdom
O. Pech

¹⁶Krankenhaus Barmherzige Brüder Regensburg, Regensburg, Germany
T. Beyna

¹⁷Evangelisches Krankenhaus Düsseldorf, Kirchfeldstraße, Düsseldorf, Germany
F. Van Der Sommen

²Eindhoven University of Technology, Eindhoven, Netherlands
P. De With

²Eindhoven University of Technology, Eindhoven, Netherlands
J. De Groof

¹Amsterdam UMC, locatie VUmc, Amsterdam, Netherlands
J.J.G.H. M. Bergman

¹Amsterdam UMC, locatie VUmc, Amsterdam, Netherlands

Abstract

Full Text

Aims Timely endoscopic detection of Barret's neoplasia has significant influence on patient outcome. Computer Aided Detection (CAD) systems may assist in neoplasia detection.

Methods The system was pretrained with ImageNet followed by domain-specific pretraining with GastroNet. GastroNet comprises >5 million endoscopic images. The system was then trained and validated on a BE dataset originating from 15 international endoscopy centers, comprising 6.337 neoplastic (1.362) patients) and 7.695 non-dysplastic images (1.139 patients). All images had histopathological confirmation. Neoplastic images were delineated by expert endoscopists. The system was tested on two prospective video test sets. The test set 1, comprising 71 neoplastic (45 cases) and 180 non-dysplastic (66 cases) videos, included all consecutive cases acquired from January to March 2022. Test set 2 comprised 47 neoplastic (47 cases) and 141 non-dysplastic (82 cases) videos and was enriched with subtle cases of neoplasia. Test set 2 was evaluated by 63 general endoscopists in without and with ADe assistance. Finally, 14 international, independent BE experts evaluated this test set.

Results Sensitivity and specificity of the CAD system were 97% and 85% for test set 1 and 91% and 82% for test set 2. Sensitivity of general endoscopists increased from 67% to 79% with CAD assistance, whilst specificity decreased from 96% to 94%. The sensitivity and specificity of experts were 86% and 90%, respectively ([Table 1]).

Table 1

Conclusions CAD outperformed general endoscopists in detecting BE neoplasia. Providing CAD to general endoscopists significantly improves their detection rate. CAD detects virtually all neoplasia in a test set representing daily practice. CAD has a detection rate on par with experts.

Table 1) Performance of general endoscopists, expert endoscopists and the CAD system on test set 2

Figures

Table 1