New Chest Imaging Database Assists AI in the COVID-19 Fight

Published today in the Open-Access, Open-Data journal GigaScience is the National COVID-19 Chest Imaging Database (NCCID), a centralized database of chest radiographs, computed tomography (CT) and MRI scans of patients in the UK. Leveraging its unique position as the world’s largest integrated healthcare system, the benefits of breast imaging data collection are so great and are already being used by physicians and the research community. The database already supports the development of artificial intelligence (AI) powered image processing software and diagnostic products and models used to predict COVID-19 death in the UK. And also has the potential to become a resource for teaching radiologists in the long term. These efforts have the potential to enable faster patient assessment in case of accidents and emergencies, save doctors time and increase the safety and consistency of care in the UK.

Representative anonymized lung image from the National COVID-19 Chest Imaging Database (NCCID). Credit: NCCID Collaborative

With the GigaScience paper describing how to access this Open Data resource, the NCCID training data is available to users worldwide, including software developers, academics and clinicians, through a rigorous data access request process. The National Covid Chest Imaging Database (NCCID) is one of the largest datasets of its kind and is constantly growing, with 25 NHS trusts and over 25 NHS trusts and 27 hospitals contributing. Collection of imaging data from over 10,000 patients in the NCCID training set to provide 34,559 imaging studies and 1,663 GB of data.

Collected with the aim of supporting a better understanding of the COVID-19 virus and developing technology that enables the best care for patients hospitalized with a serious infection. It is an initiative of NHSX, the new unit of the UK Government responsible for setting national policies and developing best practices for National Health Service (NHS) technology, digital and data, including data sharing. and transparency. This represents a nationwide effort of radiologists and IT teams at NHS sites, the British Society of Thoracic Imaging (BSTI), Royal Surrey County Hospital (RSCH) and the NHS AI Lab have worked together at a time when all teams had limited capacity for long-term projects.

Asked about the usefulness of this work, first author Dominic Cushnan says: “The NCCID database facilitates the creation of AI-enabled diagnostic tools that can augment and accelerate the clinical diagnosis of diseases (such as COVID) and also enable longitudinal studies of the effect supporting disease on patients and population management”

This data has already been used by a number of projects and studies, such as the LUCAS tool developed to predict the mortality risk of Covid patients based on patient variables (see the preprint https://doi.org/10.1101/2021.04.27.2256196 ). As well as a number of artificial intelligence-powered models and algorithms to help doctors decide what action to take on the front lines, such as administering oxygen and drugs, before patients reach a critical stage. The Cambridge University NHS AIX-COVNET collaboration is one of the first projects to use this data to develop an open-source AI tool to support the rapid diagnosis and triage of patients with COVID-19.

Pseudonymized data is provided through the NHSE AWS cloud infrastructure after completing a data access request. See the website for more: https://nhsx.github.io/covid-chest-imaging-database

Reference: “An Overview of the National COVID-19 Chest Imaging Database: Data Quality and Cohort Analysis” Nov 29, 2021, GigaScience.
DOI: 10.1093/gigascience/giab076