Deep Learning for Detection and Segmentation of Artefact and Disease Instances in Gastrointestinal Endoscopy

From LRDE

Abstract

The Endoscopy Computer Vision Challenge (EndoCV) is a crowd-sourcing initiative to address eminent problems in developing reliable computer aided detection and diagnosis endoscopy systems and suggest a pathway for clinical translation of technologies. Whilst endoscopy is a widely used diagnostic and treatment tool for hollow-organs, there are several core challenges often faced by endoscopistsmainly: 1) presence of multi-class artefacts that hinder their visual interpretation, and 2) difficulty in identifying subtle precancerous precursors and cancer abnormalities. Artefacts often affect the robustness of deep learning methods applied to the gastrointestinal tract organs as they can be confused with tissue of interest. EndoCV2020 challenges are designed to address research questions in these remits. In this paper, we present a summary of methods developed by the top 17 teams and provide an objective comparison of state-of-the-art methods and methods designed by the participants for two sub-challenges: i) artefact detection and segmentation (EAD2020), and ii) disease detection and segmentation (EDD2020). Multi-center, multi-organ, multi-class, and multi-modal clinical endoscopy datasets were compiled for both EAD2020 and EDD2020 sub-challenges. The out-of-sample generalization ability of detection algorithms was also evaluated. Whilst most teams focused on accuracy improvements, only a few methods hold credibility for clinical usability. The best performing teams provided solutions to tackle class imbalance, and variabilities in size, origin, modality and occurrences by exploring data augmentation, data fusion, and optimal class thresholding techniques.

Documents

Bibtex (lrde.bib)

@Article{	  boutry.21.media,
  author	= {Sharib Ali and Mariia Dmitrieva and Noha Ghatwary and
		  Sophia Bano and Gorkem Polat and Alptekin Temizel and
		  Adrian Krenzer and Amar Hekalo and Yun Bo Guo and Bogdan
		  Matuszewski and Mourad Gridach and Irina Voiculescu and
		  Vishnusai Yoganand and Arnav Chavan and Aryan Raj and Nhan
		  T. Nguyen and Dat Q. Tran and Le Duy Huynh and Nicolas
		  Boutry and Shahadate Rezvy and Haijian Chen and Yoon Ho
		  Choi and Anand Subramanian and Velmurugan Balasubramanian
		  and Xiaohong W. Gao and Hongyu Hu and Yusheng Liao and
		  Danail Stoyanov and Christian Daul and Stefano Realdon and
		  Renato Cannizzaro and Dominique Lamarque and Terry
		  Tran-Nguyen and Adam Bailey and Barbara Braden and James
		  East and Jens Rittscher},
  title		= {Deep Learning for Detection and Segmentation of Artefact
		  and Disease Instances in Gastrointestinal Endoscopy},
  journal	= {Medical Image Analysis},
  volume	= {70},
  number	= {102002},
  year		= {2021},
  month		= may,
  doi		= {10.1016/j.media.2021.102002},
  abstract	= {The Endoscopy Computer Vision Challenge (EndoCV) is a
		  crowd-sourcing initiative to address eminent problems in
		  developing reliable computer aided detection and diagnosis
		  endoscopy systems and suggest a pathway for clinical
		  translation of technologies. Whilst endoscopy is a widely
		  used diagnostic and treatment tool for hollow-organs, there
		  are several core challenges often faced by endoscopists,
		  mainly: 1) presence of multi-class artefacts that hinder
		  their visual interpretation, and 2) difficulty in
		  identifying subtle precancerous precursors and cancer
		  abnormalities. Artefacts often affect the robustness of
		  deep learning methods applied to the gastrointestinal tract
		  organs as they can be confused with tissue of interest.
		  EndoCV2020 challenges are designed to address research
		  questions in these remits. In this paper, we present a
		  summary of methods developed by the top 17 teams and
		  provide an objective comparison of state-of-the-art methods
		  and methods designed by the participants for two
		  sub-challenges: i) artefact detection and segmentation
		  (EAD2020), and ii) disease detection and segmentation
		  (EDD2020). Multi-center, multi-organ, multi-class, and
		  multi-modal clinical endoscopy datasets were compiled for
		  both EAD2020 and EDD2020 sub-challenges. The out-of-sample
		  generalization ability of detection algorithms was also
		  evaluated. Whilst most teams focused on accuracy
		  improvements, only a few methods hold credibility for
		  clinical usability. The best performing teams provided
		  solutions to tackle class imbalance, and variabilities in
		  size, origin, modality and occurrences by exploring data
		  augmentation, data fusion, and optimal class thresholding
		  techniques.}
}