Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

COVID-19 Image Data Collection: Prospective Predictions Are the Future

Joseph Paul Cohen, Paul Morrison, Lan Dao, Karsten Roth, Tim Q Duong, Marzyeh Ghassemi

TL;DR

This work provides the first public COVID-19 chest X-ray dataset with rich prospective metadata collected from 26 countries, enabling prognostic and management-oriented ML tasks. It emphasizes domain shift through a LOCO evaluation and offers baseline results using a DenseNet-derived feature space across diagnosis, severity, survival, and trajectory tasks. The authors highlight the potential and limitations of imaging alone for COVID-19 prognosis, and propose concrete future directions (saliency, OoD detection, and domain adaptation) to accelerate multicenter validation. Overall, the dataset and task definitions aim to catalyze rapid, cross-country ML development for COVID-19 imaging-based prognosis and care planning.

Abstract

Across the world's coronavirus disease 2019 (COVID-19) hot spots, the need to streamline patient diagnosis and management has become more pressing than ever. As one of the main imaging tools, chest X-rays (CXRs) are common, fast, non-invasive, relatively cheap, and potentially bedside to monitor the progression of the disease. This paper describes the first public COVID-19 image data collection as well as a preliminary exploration of possible use cases for the data. This dataset currently contains hundreds of frontal view X-rays and is the largest public resource for COVID-19 image and prognostic data, making it a necessary resource to develop and evaluate tools to aid in the treatment of COVID-19. It was manually aggregated from publication figures as well as various web based repositories into a machine learning (ML) friendly format with accompanying dataloader code. We collected frontal and lateral view imagery and metadata such as the time since first symptoms, intensive care unit (ICU) status, survival status, intubation status, or hospital location. We present multiple possible use cases for the data such as predicting the need for the ICU, predicting patient survival, and understanding a patient's trajectory during treatment. Data can be accessed here: https://github.com/ieee8023/covid-chestxray-dataset

COVID-19 Image Data Collection: Prospective Predictions Are the Future

TL;DR

This work provides the first public COVID-19 chest X-ray dataset with rich prospective metadata collected from 26 countries, enabling prognostic and management-oriented ML tasks. It emphasizes domain shift through a LOCO evaluation and offers baseline results using a DenseNet-derived feature space across diagnosis, severity, survival, and trajectory tasks. The authors highlight the potential and limitations of imaging alone for COVID-19 prognosis, and propose concrete future directions (saliency, OoD detection, and domain adaptation) to accelerate multicenter validation. Overall, the dataset and task definitions aim to catalyze rapid, cross-country ML development for COVID-19 imaging-based prognosis and care planning.

Abstract

Across the world's coronavirus disease 2019 (COVID-19) hot spots, the need to streamline patient diagnosis and management has become more pressing than ever. As one of the main imaging tools, chest X-rays (CXRs) are common, fast, non-invasive, relatively cheap, and potentially bedside to monitor the progression of the disease. This paper describes the first public COVID-19 image data collection as well as a preliminary exploration of possible use cases for the data. This dataset currently contains hundreds of frontal view X-rays and is the largest public resource for COVID-19 image and prognostic data, making it a necessary resource to develop and evaluate tools to aid in the treatment of COVID-19. It was manually aggregated from publication figures as well as various web based repositories into a machine learning (ML) friendly format with accompanying dataloader code. We collected frontal and lateral view imagery and metadata such as the time since first symptoms, intensive care unit (ICU) status, survival status, intubation status, or hospital location. We present multiple possible use cases for the data such as predicting the need for the ICU, predicting patient survival, and understanding a patient's trajectory during treatment. Data can be accessed here: https://github.com/ieee8023/covid-chestxray-dataset

Paper Structure

This paper contains 34 sections, 8 figures, 13 tables.

Table of Contents

  1. Introduction
  2. Background and Related Work
  3. Learning in Medical Imaging
  4. Use of Imaging for COVID-19
  5. COVID-19 Prediction from CXR
  6. Cohort Details
  7. Data Collection
  8. Experimental Setup
  9. Leave-One-Country/Continent-Out (LOCO) Evaluation:
  10. Models and Features
  11. Task Ideas with Baseline Evaluations
  12. Complement to COVID-19 Pneumonia Diagnosis and Management
  13. Motivation:
  14. Task Specification:
  15. Results:
  16. ...and 19 more sections

Figures (8)

  • Figure 1: Counts of each pneumonia frontal CXR by type and genus or species when applicable. The hierarchy structure is shown in the table from left to right. Information is collected by manually reading clinical notes for a mention of a confirmed test.
  • Figure 2: Age per patient
  • Figure 3: Offset per image
  • Figure 4: A UMAP McInnes2018UMAP visualization of each CXR from this dataset together with the Kaggle RSNA pneumonia images. CXRs with a trajectory are shown with an arrow between timepoints. If survival outcome is known the arrows and/or points are colored. The background is colored based on the density of points in the ICU or that are intubated.
  • Figure 5: Histograms of image sizes in pixels.
  • ...and 3 more figures