Enhanced high resolution spectral analysis of sleep spindles
Olivier Caspary, M. Tomczak, N. Di Renzo, Marc Mouzé-Amady, Didier Henry · 2002
A method to extract the spectral components from spindles of sleep electroencephalograms (EEG) is proposed. High resolution (HR) spectral methods are used which yield better results than Fourier transform (FT) or parametric estimators. A zoom function is produced to reduce their numerical complexity and increase their resolution. The method allows to clearly separate the different spectral components (spindles, K complexes, background activity), and should be considered for spectral analysis of various short EEG transients.