Analyse quantitative de l’électroencéphalogramme en sommeil paradoxal

Abstract

The analysis of the electroencephalogram (EEG) during paradoxical sleep (REM) is a problem in all areas requiring this analysis, whether for the study of behavioural disorder in paradoxical sleep (RBD) or the study of microstates, for example. The major problem encountered in REM remains the omnipresence of rapid eye movements (REM) in the EEG and these mask the underlying cerebral activity. This thesis offers a method to automatically correct ocular artifacts (OA) present on the EEG using independent component analysis (ICA). In addition, a semi-automatic rapid eye movement detector will be implemented in REM sleep to increase the accuracy of the corrector as well as providing an essential tool for REM analysis for RBD patients (REM Sleep Behaviour Disorder). Finally, the analysis of the existing methods for identifying microstates during REM sleep will complete this quantitative analysis. Since these characteristic topographies on the scalp are a representation of the underlying brain activity, it is essential to clean up the trace of ocular activity on the REM EEG to not bias these microstates. This thesis therefore offers an algorithm to correct while preserving theta activity and even increase it a little in REM as well as reduce the delta activity associated with eye movements. A semi-automatic detector makes it possible to make a first pass very quickly and thus greatly accelerate the detection of experts to then present the microstates in paradoxical sleep. These tools will make it possible to analyze all the paradoxical sleep, but principally discover the brain process during these microstates. All these tools will be available via the Snooz application developed at the Center for Advanced Research in Sleep Medicine (CARSM), allowing researchers and clinicians to perform digital analyzes on sleep using the tools that will be accessible. This platform will also allow interested parties to implement their own sleep tools.

Date	21 Dec 2022
Original language	French
Awarding Institution	École de technologie supérieure
Supervisor	Jean-Marc Lina (Supervisor)