I have a permanent position as "chargé de recherche" (CRHC) at the CNRS. I am particularly interested in data processing of EELS (Electron Energy Loss Spectroscopy) spectrum images. The sequences of EELS spectra acquired by scanning the electron probe over the sample constitute image spectra (also called hyperspectral images) whose size is of the order of 100x100 pixels x 1000 energy channels. It is possible to apply conventional EELS data processing techniques to each spectrum. However, it is preferable to consider the spectrum-image as a whole and to exploit the redundant nature of the information by applying multivariate statistical analysis methods. In particular, I rely on the numerous works carried out on hyperspectral images in the field of remote sensing through my collaboration with Nicolas Dobigeon's team (IRIT/INP - ENSEEIHT Toulouse http://dobigeon.perso.enseeiht.fr/ ). I was particularly involved in denoising, reconstruction of partially sampled spectrum-images and unsupervised classification (clustering).