Evaluation of a Filter/Integrator System for Quantifying Delta Activity in the Sleep Electroencephalogram

Abstract

Summary: The integrated output of a delta bandpass filter was evaluated as a means of quantifying delta activity in the sleep electroencephalogram (EEG). Total integrated slow wave amplitude (ISWA) and a density measure (ISWAI 20-s epoch) were determined from four complete NREM sleep periods ,,(NREMPs) on each of'2 base.Jine·nightsfrom15"subjects. T@tallSWAdisplays a linear decline across NREMPs, whereas ISWAI20-s epoch shows a curvilinear decline. Both measures show highly significant correlations with manually staged slow wave sleep and a high degree of intra-individual consistency between nights, for both whole-night values and those of individual NREMPs. The slow-wave density index (ISWA/20-s epoch) demonstrated consistently higher between-night correlations across NREMPs than did the total ISWA index, particularly in the first NREMP, suggesting it may better characterize the nightly distribution of delta activity in an individual's sleep EEG. Key Words: Automatic scoring of sleep-Delta sleep-Slow wave sleep.Current views of sleep regulation (1,2) suggest that NREM sleep is a qualitatively homogeneous state, which varies on an intensity dimension reflected by the amplitude and incidence of delta activity in the sleep electroencephalogram (EEG). Manual determination of the proportion of sleep stages 2, 3, and 4 is subjective and inherently limited in its quantification of delta activity.Recognizing the need for more precise quantification of delta activity in sleep, some investigators have applied spectral analysis [e.g., (I)] or period and amplitude analysis (PAN) (3,4) to the sleep EEG. Both techniques produce reliable data and provide better resolution of delta activity than manual methods. However, the expense and complexity of such computer-based systems has precluded their widespread use. An alternative approach to quantifying sleep EEG consisted of simply integrating the unfiltered EEG (5-7), a technique that has rarely been applied in subsequent sleep studies. In our laboratory we have refined this approach by integrating the EEG output of a delta bandpass filter, thereby offering a simple means of quantifying delta sleep without a computer. Here we describe the operating characteristics of the system, evaluate its reliability, and present comparisons with manually staged sleep.