Brain Research Bulletin 63 (2004) 399–405 Quantitative analysis of sleep EEG microstructure in the time–frequency domain Fabrizio De Carli a,* , Lino Nobili b , Manolo Beelke b , Tsuyoshi Watanabe b , Arianna Smerieri c , Liborio Parrino c , Mario Giovanni Terzano c , Franco Ferrillo a,b a Istituto di Bioimmagini e Fisiologia Molecolare, CNR, Via de Toni 5, 16132 Genova, Italy b Department of Motor Science and Rehabilitation, Sleep Disorders Center, University of Genoa, Genoa, Italy c Department of Neurology, Sleep Disorders Center, University of Parma, Parma, Italy Available online 19 May 2004 Abstract A number of phasic events influence sleep quality and sleep macrostructure. The detection of arousals and the analysis of cyclic alternating patterns (CAP) support the evaluation of sleep fragmentation and instability. Sixteen polygraphic overnight recordings were visually inspected for conventional Rechtscaffen and Kales scoring, while arousals were detected following the criteria of the American Sleep Disorders Association (ASDA). Three electroencephalograph (EEG) segments were associated to each event, corresponding to background activity, pre-arousal period and arousal. The study was supplemented by the analysis of time–frequency distribution of EEG within each subtype of phase A in the CAP. The arousals were characterized by the increase of alpha and beta power with regard to background. Within NREM sleep most of the arousals were preceded by a transient increase of delta power. The time–frequency evolution of the phase A of the CAP sequence showed a strong prevalence of delta activity during the whole A1, but high amplitude delta waves were found also in the first 2/3 s of A2 and A3, followed by desynchronization. Our results underline the strict relationship between the ASDA arousals, and the subtype A2 and A3 within the CAP: in both the association between a short sequence of transient slow waves and the successive increase of frequency and decrease of amplitude characterizes the arousal response. © 2004 Elsevier Inc. All rights reserved. Keywords: Arousal; Cyclic alternating pattern; Wavelet analysis; EEG synchronization 1. Introduction A number of phasic events and periodic patterns make up sleep microstructure influencing sleep quality and the structure of NREM–REM sleep cycles. In the last years the evaluation of sleep has developed in particular by the detec- tion of arousals, considered as major markers of sleep frag- mentation, and by the analysis of cyclic alternating patterns (CAP), considered as indicators of sleep instability. According to the American Sleep Disorders Associa- tion (ASDA), electroencephalographic (EEG) arousals are transient events characterized by “an abrupt shift in EEG * Corresponding author. Tel.: +39-010-3537466; fax: +39-010-354180. E-mail address: fabrizio@dism.unige.it (F.D. Carli). frequency, which may include theta, alpha and/or frequen- cies greater than 16 Hz but not spindles” [1]. A certain number of spontaneous arousals seem to be an intrinsic component of physiological sleep [4,11], but in case of sleep disorders they may become very frequent and associated with daytime sleepiness [8,15]. In some cases exogenous or endogenous stimuli pro- moting arousals may be identified whereas in other cases arousals may only be associated to variations of arousal level. Transient changes in EEG have been observed in asso- ciation with vegetative modifications, such as modifications in heart rate and blood pressure: these EEG changes may be characterized by different patterns, including arousals, K-complexes and delta bursts. Such transient EEG patterns, emerging from the background activity, have often been 0361-9230/$ – see front matter © 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.brainresbull.2003.12.013