Nikos Paragios: a life devoted to artificial vision and its applications in medicine

Changed on 25/03/2020
This year, eight ERC grant winners are leaders or members of Inria project-teams. Nikos Paragios, a teacher at the Ecole Centrale de Paris, was awarded a starting grant. Worth approximately 1,500,000 euros, this grant will allow the researcher to reinforce the GALEN team and conduct an ambitious five-year research project on artificial vision and its applications in the medical field. Here, we get to know the grant-winning scientist.

After a brief stay in Rennes, Nikos Paragios, originally from a small island in the Aegean Sea, returned to his Mediterranean roots.  Between 1996 and 1999, he did his PhD on artificial vision at the Inria Sophia-Antipolis research centre, before working at the Siemens research centre at Princeton (USA), while also teaching at New York University and the State University of New Jersey. After five years of research, during which he filed around thirty patents based around medical imaging, he felt the need to return to Europe. He taught at the Ecole des Ponts et Chaussées and the Ecole Centrale de Paris, then, in 2008, took charge of the GALEN team, a joint team involving the Inria Saclay - Ile-de-France research centre and the Ecole Centrale, whose work concerns the modelling of organs through the extraction, representation and understanding of the content of medical images. One common theme throughout his career has been the applications of artificial vision in the medical field, work that won him an ERC starting grant to consolidate his team.

For Nikos Paragios, the social impact of his research is of paramount importance: "if it succeeds, it can make an enormous difference to everyday life," he says about his work to replicate human vision. Vision is a function that monopolises a large part of the brain and represents an enormous challenge for the researcher given that even the most powerful computer struggles to reproduce even a small fraction of a human being's visual capabilities. "It is capable of only a tiny fraction of what a 5-year-old child can do ." Among the reasons why he chose to pursue this discipline was its cross-disciplinary nature: "it touches upon applied mathematics, medicine, biology, computer science and more... We are trying to create a unifying theoretical approach and an accompanying software platform that is capable of reproducing intelligent systems that resemble biological systems. The aim is to help doctors to make early diagnoses based on automated processing of the data from medical imaging (scans, MRI, etc.) ".

Computers will act as the doctor's eyes... This will make it possible to see the evolution of a tumour over time and to spot illnesses early.

His research is conducted in a context in which recent technological developments have given rise to a new generation of scanners, as well as new acquisition methods which enable in vivo visualisation of the anatomical structures of biological systems in a non-invasive way. The data from these medical imaging tools cannot be interpreted with the naked eye. At least not yet. Thanks to DIOCLES, the software tool developed by Nikos Paragios, "Computers will act as the doctor's eyes, assisting them in analysing data from scans, MRIs, etc. , " he explains. " This will make it possible, for example, to see the evolution of a tumour over time or spot an illness early ". But, instead of creating a different approach for each illness, as is the case today (for heart attacks, for instance), the researcher is focusing on a generic platform which makes it possible, when processing any image, to deal with the greatest possible number of illnesses. 

For the GALEN team, the aim is to use discrete optimisation methods and statistical models in order to identify normal cases and pathological cases and compare them to those of a given patient, while trying to provide suitable responses to a wide range of variations: Is the person elderly? What is their medical history?, etc.

The modelling and understanding of the long-term effects of ageing is of great importance for many organs and illnesses which do not have any pre-clinical indicators, such as neurological diseases of the brain, muscular diseases and some forms of cancer. This research is, to a large extent, done in partnership with hospitals (CHU Henri Mondor de Créteil, Pitié Salpêtrière), which provide the team with the necessary data and knowledge relating to the types of illness being studied. "There are constant and very productive exchanges  between us," says the researcher, who is delighted to be receiving this grant for the collaboration between the Ecole Centrale de Paris and Inria, a partnership between engineers who are capable of tackling the major challenges of the 21st century and excellent scientists assisted by top-class support staff.

He had already chosen the institute as the place to further his studies after leaving the University of Crete, attracted by its "exceptional international profile and its reputation for quality". The ERC grant will allow him to recruit "young professors or lecturers in applied mathematics and computer science to help me strengthen the theoretical basis required to build this platform ". This goal does not worry Nikos Paragios.

It's quite an open field and it's expanding fast; there are already a lot of researchers working on this subject in France and across the world and it offers interesting career development prospects. And it is always easier to motivate people to get involved with risky, cutting-edge research with a social dimension.