My first experience with Inria was back when I did my thesis, following a Masters in applied mathematics. That was when I made a life-changing encounter: Professor Alain Trouvé, who led me to decide to commit to a research career. I carried out my thesis in the Asclepios project team led by Nicholas Ayache, who introduced me to the fascinating world of automatic analysis of medical images. I then went to the United States as a post-doc, with Professor Guido Gerig at the Scientific Computing and Imaging Institute, where I continued my research on the exploitation of repeated observations over time. We call them "longitudinal data". This subject became one of my favourite themes. I then returned to Inria, where I was seconded to Olivier Colliot's ARAMIS team at the ICM in Paris. With my arrival, the ARAMIS team became a joint team with Inria.
My research consists of constructing digital models of the structure and function of the brain. These models are built thanks to computer programs that average the information contained in the medical images of a group of subjects or patients. Together with my PhD students and post-docs, we develop new statistical tools to analyse this neuroimaging data, which is particularly complex: made up of images, surfaces and curves that represent different aspects of the anatomy and functioning of the brain. There are numerous applications for these digital models: we can visualise the effects of a pathology on the brain and therefore help clinicians to make assumptions on the biological mechanisms of these pathologies. By comparing a patient's data with the models, we are going to be able to produce tools to assist clinicians with the diagnosis and prognosis of neuro-degenerative diseases (Alzheimer, Parkinson). These pathologies are often diagnosed late, because the first symptoms appear at an advanced stage of the disease, following a long silent phase. At the ICM - located at the site of the Pitié Salpêtrière hospital - we are ideally placed to carry out this research interacting with our hospital partners and reference centres for numerous neurological diseases.
I find the mathematical modelling of complex problems very exciting. And, in terms of complexity, the brain provides an unequalled playing field! Today, medical imaging allows us to observe the brainin vivoin its normal or pathological functioning. This data provides a mass of information that is essential in order to understand the brain and neurological diseases. However this data is so complex that we can only extract pertinent information with the help of increasingly sophisticated mathematical, statistical and computing tools. During my career, I have studied databases of longitudinal cerebral images: these are series of pictures of the brain of several individuals, taken at different moments in their lives. Using this data, my aim is to reveal constants in the brain development or ageing. I have developed my own statistical and algorithmic tools in order to reveal, within a sample of individuals, an average shape and its variations. Within this context, I have developed a methodology capable of establishing a scenario for the average growth of the amygdala (a deep structure in the brain), in children. This has enabled me to bring to light earlier amygdala growth in autistic children.
Firstly, I would like to expand my team by recruiting at least three PhD students and one or two post-docs. Our first job will be to develop a new theoretical framework in order to build dynamic models of the brain during ageing or during the progression of a disease. In order to establish average scenarios from observations of individuals repeated over time, we need to compare what is comparable. Between two individuals, the same pathology does not emerge at the same time and does not evolve at the same pace. The age criterion is therefore ineffective. We are going to have to invent new statistical tools that will automatically match similar events (such as the appearance of a lesion in a certain area of the brain, for example) that happen at different times for each patient, thereby replacing the disease in the life and history of the patient.
