PARIS (TICpharma) – While the Paris Public Hospital Assistance (AP-HP) announced in June a collaboration with GE Healthcare to create a chest imaging database of patients suspected of Covid-19, the French Society of Radiology (SFR) also announced at the French Radiology Days (JFR) in spring that it had set up a database of chest scanners to be treated with artificial intelligence.
AP-HP announced in a statement on June 16 that it was working with GE Healthcare, Orange Healthcare and TheraPanacea “to conduct a research project to create a database” of chest imaging in patients suspected of Covid-19.
The project called “Stoic” for “thoracic scanning for the diagnosis of pneumonia related to Covid-19”, aims to analyze 10,000 chest scanners carried out for suspected Covid-19. It is worn by Professor Marie-Pierre Revel, Head of the Cardiothoracic Imaging Unit at Cochin Hospital.
“The focus of the scanner is to identify the signs of the disease early and assess its extent,” Prof Revel said in the AP-HP statement. “But it also provides further information on the patient’s state of health, pre-existing to the infection, allowing to establish a score of severity that can be correlated to clinical evolution. All this data is now available and should lead to a better understanding of why some individuals develop a severe form of the disease.”
The project involved a group of 20 radiologists, experts from the Thoracic Imaging Society. They employ a 3D image visualization application, developed by GE Healthcare on the Edison platform and deployed by Orange Healthcare through secure network access and healthcare data hosting infrastructure.
This software allows you to view 3D images and remotely annotate lung damage by bypassing areas affected by the virus. It also provides information on factors of vascular, pulmonary or overweight comorbidity that may influence the course of the disease such as the appearance of the arteries, that of the unaffected lungs, or the abundance of fat in the chest wall.
This database will enable the development of artificial intelligence solutions, for example, to quickly and automatically quantify the extent and severity of lesions, to guide patient management, or to evaluate the effectiveness of the treatments implemented.
Medical technology company TheraPanacea has developed a pilot incorporating artificial intelligence to quantify the disease and establish a short-term prognosis of patients from their hospital entry lung scanners.
“Artificial intelligence has a crucial role to play in crisis situations,” notes Professor Nikos Paragios, President of TheraPanacea. It allows doctors and caregivers to “make the right decisions faster and for more patients, while keeping their critical mind”.”
“This health crisis has forced us to mobilize fully and urgently, and no one has counted their time, be it engineers, radiologists, researchers, or project managers managing the regulatory aspects,” said Prof. Revel.
All pseudo-anonymized data comes from many public hospitals in Paris, but also from Rennes, Lyon and Strasbourg.
Finally, “Stoic” is supported by a sponsorship of several companies and foundations: Innothera, Guerbet, the Central Supélec Foundation and the AP-HP Foundation for Research.
The Fidac-SFR database
Another project to collect and analyze chest scanners of Covid patients who have been proven or suspected, the Fidac project (for french imaging database against coronavirus)was developed by the French Society of Radiology (SFR) in partnership with Nehs Digital.
The SFR was requested by “DRIM France IA”, the first French artificial intelligence ecosystem for medical imaging, to be the promoter of ananonymized database. “The aim is to quickly make available a set of data from different centres according to a standardized acquisition protocol (preferably intra-injection transverse cut) so that researchers can develop segmentation, detection, classification, gravity scores to better manage the disease,” the SFR says on its website.
The Fidac base “aims to develop algorithms, free of law, that can help radiologists diagnose, quantify, assess the severity, the prognosis of patients,” says Prof. Jean-Paul Beregi of the Nîmes University Hospital, who is at the initiative of this database.
Nehs Digital is assisting in setting up the national collection and provides, among other things, a centralized portal and health data hosting “for the time of the partnership and for the operation of this database”.
To date, the Fidac database contains more than 3,000 files, Jean-François Meder, president of the SFR, told APMnews (the website of the APM International group, which includes TICpharma). And “we have received many requests to use this database,” he added.
The aim is to be able to offer institutional or private researchers, under the control of the SFR and “DRIM France IA” data to identify diagnostic biomarkers, prognostics, particular forms, analyze follow-up, correlate radiological data to the PCR, etc.
“For example, this type of tool can identify complications of Covid-19 disease that were initially unknown, such as thromboembolic diseases,” added the SFR president.
“The AP-HP Stoic project and the Fidac-SFR project are quite identical,” confirmed Jean-François Meder. “Technically, the merger of the two databases would be possible and personally, it seems to me to be extremely desirable,” he concluded.
Geneviève De Lacour