- AI-based solution will analyse lung scan imagery to provide initial diagnosis for healthcare professionals.
- Developed for a French defence ministry call for projects launched by the Defence Innovation Agency with a 668,000 euros funding, the solution will be ready for deployment in the autumn of 2020.
The COVID-19 pandemic has put huge strain on healthcare services around the world, and this situation will likely continue for many months in some countries. To prevent hospital services becoming overstretched and better prepare for dealing with this kind of crisis in the future, the French Society of Radiology (Société Française de Radiologie – SFR) recommends CT (computed tomography) scans combined with mobile X-ray to diagnose COVID-19. These scans are performed by a radiologist, who has to interpret multiple images for each patient, often within a short timeframe, in order to assess lung damage.
Thales and NEHS DIGITAL, both major players in the French healthcare system, have joined forces to speed up the hospital admissions process in a crisis by using new technology. This project will leverage Thales’s expertise in radiology and artificial intelligence (AI) and NEHS DIGITAL’s extensive footprint in the French hospital system. NEHS DIGITAL is a subsidiary of nehs group.
The partners will develop a solution in record time using AI to analyse CT imagery of the chest and lungs. As soon as the images are taken, the AI service will make an initial recommendation, including a preliminary diagnosis and assessment of the level of criticality of any lung damage, so that medical staff can admit the patient to the appropriate ward and prioritise the most urgent cases. The project was selected after a call for projects launched by the French Ministry for the Armed Forces to combat COVID-19.
Databases of medical imagery relating to the COVID-19 epidemic are virtually non-existent today. NEHS DIGITAL has now begun collecting anonymised chest scan data from around 100 hospitals in France as part of the FIDAC (French Imaging Database Against Coronavirus) project (1). To analyse these datasets, which must be as large as possible to ensure they are representative, Thales will set up an infrastructure to train the necessary algorithms using machine learning and develop an AI system that can make automated recommendations to healthcare professionals.
Deployment of the AI service developed by Thales will make use of NEHS DIGITAL’s extensive cybersecure telemedicine infrastructure Hosted on a secure cloud. The partners aim to deploy an initial demonstrator of this solution over the next three months using NEHS DIGITAL’s existing platform, which is already used by almost all French hospitals. The solution will be continuously updated as the number of available images increases.
A medical committee will validate the solution and monitor rollout in hospitals.
This large-scale project is a first step towards more widespread use of AI to support the work of radiologists. Thanks to this collaborative approach by multiple stakeholders, and with the architecture already in place, the AI service could also help to diagnose other pathologies in the future.
(1) FIDAC, French Imaging Database Against Coronavirus: The French Society of Radiology (SFR), in partnership with NEHS DIGITAL as well as the CERF and DRIM France IA, has developed a national database of chest scanners with the aim to improve patients care that are positive to Covid-19. The project aims to build an anonymized database of chest CT images from patients suspected or affected by COVID-19 to develop detection and classification tools in order to better manage patients with SARS-Cov2.