HINT

Medical imaging provides a wide spectrum of modalities giving access to data of various physical natures and dimensions. Each imaging modality provides valuable information for the diagnosis and monitoring of patients. Nevertheless, the linking of these information is made complex by the heterogeneity between these data. This heterogeneity is further increased by differences in equipment and acquisition protocols, especially in the case of multicenter studies. In neonatology, and more particularly in the services taking care of premature newborns, imaging technologies are essential, both for the daily treatment of patients and for research aiming at understanding the mechanisms of functional maturation of the brain, and thus developing new care protocols. Nevertheless, there are few appropriate tools for the use of source data for such studies (Magnetic Resonance Imaging - MRI, UltraSound - US, ElectroEncephaloGraphy - EEG). Our goal is to develop numerical methods and software tools allowing clinicians to study neurodevelopmental trajectories in neo- and perinatology. However, in this context, the available data are much less numerous than in adults, adding to the problems of heterogeneity difficulties related to the low volume of data sets, likely to defeat the usual paradigms of information extraction. We address these issues from several angles: the mapping of multiphysical, multidimensional heterogeneous data (3D MRI vs. 2D ultrasound, multicentric); the assistance to the analysis of signal information (EEG) and imaging by AI approaches in a heterogeneous "small data" context; the leveraging of the coupling image / signal, by the development of tools for multiphysical visualization. These tools will be involved in the processing of different neonatal data cohorts (CALIN, Epirmex, Lytonepal).