The Institut Pasteur's Biological Image Analysis Unit is taking part in a new project selected by the EU's Innovative Medicines Initiative. The project, BIGPICTURE, aims to speed up the development of artificial intelligence (AI) in medicine, especially in biomedical image analysis.
A new consortium selected by the EU's Innovative Medicines Initiative (IMI) intends to develop the biggest database of pathology images to accelerate the development of artificial intelligence in medicine.
The BIGPICTURE project aims to develop artificial intelligence (AI) applications in anatomic pathology, a medical specialty that involves observing biopsies and evaluating anomalies associated with disease in cells, tissues and organs. The BIGPICTURE project will investigate a range of diseases and medical situations including cancer, autoimmune diseases and transplant follow-up. In these situations, medical images are digitized, shared and processed by computers. This paves the way for AI applications to help scientists study diseases, stratify patients and find more effective treatments. AI applications can be used for the automatic analysis of biomedical images once they have been trained to classify images associated with results (true/false-type responses). Their ability to correctly analyze these images involves a learning phase which often requires human intervention, or manual annotation – in other words a stage in which the algorithms are taught to recognize what is true or false.
"Manual annotation is the bottleneck for many projects based on AI," notes Jean-Christophe Olivo-Marin, Head of the Institut Pasteur's Biological Image Analysis Unit, which is a member of the BIGPICTURE project consortium. One of the unit's tasks will be to create automatic annotation algorithms that will be used to set up an open distributed software platform to read virtual slides. Using algorithms based on AI for this learning phase will improve the speed and precision of the process.
The BIGPICTURE consortium includes hospitals, academic partners and pharmaceutical companies as well as various public and private organizations from 12 European countries. The project began on March 1, 2021 and will run for six years. But the aim is for the platform to last longer than this, and the consortium will draw up plans to maintain and continue to develop it beyond this period.