top of page
Search
  • Victor

DESKi & Bordeaux University Hospital Sign a Partnership to Develop AI Research in Medical Imaging

Updated: May 28, 2019

A framework agreement between DESKi and Bordeaux university hospital has been announced on May 14th. The initial project will focus on machine learning tools to democratize cardiac ultrasound imaging.

From left to right: Professor Stéphane Lafitte, Philippe Vigouroux, Managing Director of Bordeaux University Hospital (CHU), Dr. Bertrand Moal, co-founder of DESKi

Ultrasound imaging, a public health challenge

In developed countries, the need for medical imaging is growing exponentially, both for diagnostic purposes and follow-up care. Yet two thirds of the world's population do not have access to any medical imaging and the supply has stagnated since training doctors is time-consuming and cost-intensive for healthcare systems.

Against this backdrop, DESKi focuses especially on ultrasound imaging, a modality that could provide a number of solutions to address these public health issues. Ultrasound imaging is non-invasive, multi-organ, and affordable compared to other types of imaging. These aspects, combined with the recent miniaturization of probes - which are connected directly to tablets or smartphones - suggest that in the future, the doctor's stethoscope will be replaced by an ultrasound probe.

But training still remains an obstacle to this democratization; extensive experience is required to learn how to acquire and analyze ultrasound images. DESKi has therefore developed innovative solutions based on artificial intelligence to simplify the use of ultrasound systems by doctors who are not specialists.



Bordeaux University Hospital and DESKi side by side

Bordeaux University Hospital and DESKi have decided to enter into a framework agreement enabling the development of artificial intelligence projects based on fast, preferential and regular access to Bordeaux University Hospital's research facilities.

DESKi has installed a high-performance computing server dedicated to artificial intelligence at Bordeaux University Hospital in order to analyze anonymised data, in accordance with the regulations and the requirements of the French Data Protection Authority (CNIL).

"This partnership is a real opportunity for DESKi: having access to an extensive database of verified information and support from renowned medical specialists provides us with optimal conditions for developing our solutions," says Bertrand Moal.

"This partnership shows that Bordeaux University Hospital is fully involved in a new era of artificial intelligence and big data processing. Data analysis, performed in-house by Bordeaux University Hospital professionals with dual expertise in the medical and engineering fields, allows us to provide better follow-up care for patients, and better training for students and medical professionals," says Philippe Vigouroux, Managing Director of Bordeaux University Hospital.



Solution for acquisition assistance: the algorithm helps doctors correctly positon the probe in real time, in order to obtain the best possible ultrasound cross-section

Over 17,000 cardiac ultrasound exams are carried out and analyzed by cardiologists every year at Bordeaux University Hospital. DESKi is using this data for its deep learning algorithms trained at reproducing these experts’ analysis and then spread to a wider medical community.

In cooperation with Bordeaux University Hospital's Cardiology and Valvular Heart Disease Department, directed by Professor Stéphane Lafitte (unit 1034 INSERM), the project's initial goal is to develop tools to assist with cardiac analysis. DESKi then plans to expand its tools to other specialties such as gynaecology, gastroenterology and urology.

This democratization will have a great impact in developed countries like France, where medical deserts are a problem, but it will have an even bigger impact in developing countries, where healthcare and medical imaging systems have yet to be established.

248 views0 comments

Recent Posts

See All
bottom of page