A French consortium to accelerate AI development in imaging
Advances in medical imaging constantly increase our knowledge of diseases and their treatments, creating a surge in the amount of data generated. For each patient to benefit from the promise of personalized medicine, new AI-based tools are needed to aggregate, standardize and make sense of this data quickly.
However, the development of AI applications in medical imaging requires access to large amounts of high-quality annotated data, advanced visualization capabilities design and the ability to trace the developments and learnings. This can be a challenge and represent heavy investments, particularly for smaller AI players.
A French consortium of major players in AI and healthcare, which has just obtained the support of the “Future Investments” Program (PIA) led by the French General Secretary for Investment and operated by Bpifrance, aims to solve just that.
The AI DReAM project aims at industrializing the development of AI applications for imaging to accelerate their deployment to benefit patients. Led by GE Healthcare, the consortium brings together SME Evolucare, the start-ups Therapanacea and Pixyl, the Curie Institute, French Paris group of hospital AP-HP, leading European Cancer Center Gustave Roussy and the Paris Hospital Saint Joseph Foundation.
“Our aim is to create the tools that will facilitate large-scale data annotation and algorithm design, testing and validation to cut cycle development time for AI start-ups, researchers or radiologists who wish to create AI applications for medical imaging,” explains Baptiste Perrin, AI DReAM project leader at GE Healthcare.
Within this project, GE Healthcare will provide its partners with an integrated advanced visualization platform, with 2D/3D annotation tools to quickly analyze and annotate images generated by medical imaging equipment, a Software Development Kit (SDK) to develop clinical and therapeutic applications or research applications, and a technology framework that meets regulatory requirements for the development of medical applications and ensures transparency of learning data, robust learning and traceability of decision-making. The AI.DReAM platform will focus on three areas of precision medicine: early diagnosis, targeting of therapies through stratification and efficiency in the patient care pathway.
For early diagnosis, for example, members of the consortium are working on the treatment of primary liver cancer with the aim of improving survival prognosis by 15% through earlier diagnosis and earlier management of the pathology.
The aim is to develop a clinical application using AI to detect a diseased liver and automatically characterize the lesions visible in imaging (nodules of hepatocellular carcinoma), less than 3 centimeters.
Segmentation of a liver in 3D with annotation markers
“The ability to quickly and automatically identify small HCC nodules will significantly improve the management of this disease and it will be critical to the survival prognosis of patients. AI technologies can save us a lot of time and allow us to initiate treatments at the early stage of the disease,” said Professor Valérie Vilgrain, Head of Radiology at Beaujon Hospital in Paris.
The collaboration with a variety of complementary players will help accelerate the structuring of an artificial intelligence industry in healthcare in France in an extremely competitive and dynamic market.
The implementation of use cases with the partners will ensure that the platform is adequate for more projects in multiple clinical areas. Eventually, it will be offered to other partners using medical imaging.
By promoting the industrialization of AI solutions, the structuring of an artificial intelligence industry in health and the development of a business model, AI DReAM strengthens the competitiveness of France and Europe in the field of AI, in an extremely competitive and dynamic market.
This project will not only enable French SMEs and ETIs to actively participate in the international influence of artificial intelligence in healthcare, but also to take part in the ethical research on its use, and more specifically in medical imaging.