Representation of SEEG signals using time-frequency signatures

This work focuses on the analysis of stereoelectroencephalographic (SEEG) signals recorded with depth electrodes on epileptic patients. This analysis provides major information on interactions between brain structures during seizures. Considering the non stationary nature of these signals, time-frequency representations are appropriate for this analysis. The study of the time-frequency contents of SEEG signals allows to extract classes of time-frequency signatures that are characteristic of different phases of a seizure. Even if the considered classes do not form a complete dictionary of signatures, they allow to describe the evolution of the seizures.