Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Endoscopy disease detection challenge 2020

Sharib Ali, Noha Ghatwary, Barbara Braden, Dominique Lamarque, Adam Bailey, Stefano Realdon, Renato Cannizzaro, Jens Rittscher, Christian Daul, James East

TL;DR

The paper presents the EDD2020 challenge and a multi-center endoscopy dataset to address generalization in disease detection and segmentation. It defines detection, segmentation, and out-of-sample tasks with standardized metrics (IoU, mAP, Dice/F1/F2) and an out-of-sample generalization score, releasing tools and test data for benchmarking. Results on the test set show moderate overall mAP due to class imbalance, but ensemble methods improve performance, underscoring the need for robust transferability-aware approaches. The dataset and challenge provide a platform to develop and compare clinically applicable, real-time endoscopy analysis methods across institutions.

Abstract

Whilst many technologies are built around endoscopy, there is a need to have a comprehensive dataset collected from multiple centers to address the generalization issues with most deep learning frameworks. What could be more important than disease detection and localization? Through our extensive network of clinical and computational experts, we have collected, curated and annotated gastrointestinal endoscopy video frames. We have released this dataset and have launched disease detection and segmentation challenge EDD2020 https://edd2020.grand-challenge.org to address the limitations and explore new directions. EDD2020 is a crowd sourcing initiative to test the feasibility of recent deep learning methods and to promote research for building robust technologies. In this paper, we provide an overview of the EDD2020 dataset, challenge tasks, evaluation strategies and a short summary of results on test data. A detailed paper will be drafted after the challenge workshop with more detailed analysis of the results.

Endoscopy disease detection challenge 2020

TL;DR

The paper presents the EDD2020 challenge and a multi-center endoscopy dataset to address generalization in disease detection and segmentation. It defines detection, segmentation, and out-of-sample tasks with standardized metrics (IoU, mAP, Dice/F1/F2) and an out-of-sample generalization score, releasing tools and test data for benchmarking. Results on the test set show moderate overall mAP due to class imbalance, but ensemble methods improve performance, underscoring the need for robust transferability-aware approaches. The dataset and challenge provide a platform to develop and compare clinically applicable, real-time endoscopy analysis methods across institutions.

Abstract

Whilst many technologies are built around endoscopy, there is a need to have a comprehensive dataset collected from multiple centers to address the generalization issues with most deep learning frameworks. What could be more important than disease detection and localization? Through our extensive network of clinical and computational experts, we have collected, curated and annotated gastrointestinal endoscopy video frames. We have released this dataset and have launched disease detection and segmentation challenge EDD2020 https://edd2020.grand-challenge.org to address the limitations and explore new directions. EDD2020 is a crowd sourcing initiative to test the feasibility of recent deep learning methods and to promote research for building robust technologies. In this paper, we provide an overview of the EDD2020 dataset, challenge tasks, evaluation strategies and a short summary of results on test data. A detailed paper will be drafted after the challenge workshop with more detailed analysis of the results.

Paper Structure

This paper contains 9 sections, 3 equations, 2 figures.

Table of Contents

  1. INTRODUCTION
  2. Proposed tasks
  3. Dataset
  4. Evaluation
  5. Detection score
  6. Semantic segmentation score
  7. Out-of-sample detection score
  8. Analysis on EDD2020 test data
  9. Conclusion

Figures (2)

  • Figure 1: Endoscopy disease detection and segmentation dataset (EDD2020). The top two rows consists of mostly polyps, middle two rows consists of mostly suspected and high-grade dysplastic (HGD) cases in stomach, and finally bottom two row represents non-dysplastic Barrett's oesophagus (NDBE), suspected dysplasia and one case for cancer.
  • Figure 2: Detectection and localisation of disease classes on EDD2020 test dataset. A) Consists of correctly detected bounding boxes for all cases in the test data using our in-house Algorithm 1. It can be observed that The algorithm detects precise bounding boxes for non-dysplastic Barrett's (NDBE), high-grade and suspected (dysplasia). In the 2nd row, different polyps of various size and locations are localised. B) Algorithm 1 fails to detect cancer instead it confuses with polyp like protrusion. Similarly, a small Barrett's island is confused with polyp. However, this is corrected by algorithm 2 and 3.