Quantitative Description of Sleep Stage Electrophysiology Using Digital Period Analytic Techniques

Abstract

Summary:The purpose of the current study was to assess the overlap in variance of two procedures for the quantification of sleep electrophysiology: conventional stage scoring and computer quantification of tonic activity. Data were collected on 24 nights from eight subjects and were scored according to a modified set of Rechtschaffen and Kales criteria and submitted to a period-analytic computer analysis. Following this, discriminant function analyses were performed on the data for each night to predict the visual stage scores from the computer-generated data. The results indicate a very high degree of predictive accuracy (91.05%) supporting the contention that the computer-quantified data set includes the variance normally captured by stage scoring. The implications of computer quantification of sleep electrophysiology are discussed. Key Words: Sleep stagesPeriod analysis-Electroencephalography-Psychophysiology.Traditional methodologies used for the study of sleep electrophysiology and psychophysiology have principally concerned the reduction of data to stages so that the vast amounts of data collected can be reduced to a point where statistical comparisons between nights and/or subjects can be made. Historically these quantification procedures have relied on visual examination of the data and have thus emphasized easily observable visual events such as spindles, K-complexes, electromyographic (EMG) reduction, and electroencephalographic (EEG) desynchronization (1). However, recent advances in computer technology and application have suggested new ways in which such data sets can be evaluated. One direction this work has followed has been the development of machinery to automate the stage scoring process or to detect relevant phasic events (2)(3)(4)(5)(6)(7)(8).A different approach uses computers to quantify the electrophysiological data to explore psychophysiological variations within stage as well as comparing different nights or subjects using techniques other than stage scoring (9)(10)(11)(12)(13)(14)(15). In some cases these later endeavors do not directly quantify phasic events but rather include them in evaluations of tonic activity.

Quantitative EEG Analysis of Sleep Chronophysiology: A Comparison Between Root Mean Square (RMS) Estimation And Fast Fourier Transformation (FFT)

“…There still is no standardized methodology for the application of quantitative electroencephalographic (QEEG) techniques in sleep research Hasten (1985) Computer programs for data collection and analysis vary from one sleep laboratory to the next, sometimes making between-laboratory comparisons a difficult process Karacan (1978). Hoffman (1979) states that the development of new laboratory procedures (i.e. automated QEEG analysis) should be accompanied by some careful, formal discussion of the procedures.…”

“…During the past 30 years of sleep research, specific efforts have been directed towards developing automated methodologies to quantify nocturnal EEG data Agnew (1967); Armitage (1989); Armitage (1992a); Armitage (1992b); Church (1975); Dumermuth (1983): Hoffmann (1984); Johnson (1972); Kapfhammer (1992): Ktonas (1987); Ktonas (1981); Larsen (1992); Terstegge (1993); Uchida (1992a); Uchida (1992b). Much of this work has focused on formulating mathematical models that reliably characterize the two most salient features of sleep EEG (i.e., frequency and amplitude).…”

“…Uchida (1992b) reported a discrepancy between the FFT-analyzed delta-beta sleep EEG dynamics and DPA-analyzed delta-beta dynamics published by Armitage (1992a). Whereas Uchida (1992b) reported delta (0.3-3 Hz) and beta (20)(21)(22)(23)(24)(25)(26)(27)(28) to oscillate inversely across REM/NPvEM sleep, Armitage (1992a) found delta (0.5-4 Hz) and beta (16)(17)(18)(19)(20)(21)(22)(23)(24)(25)(26)(27)(28)(29)(30)(31)(32) to oscillate in phase across the sleep. In an attempt to resolve the discrepancy, Uchida (1992b) applied DPA to their data and found similar results to those yielded by the FFT.…”

On the Use of Neural Network Techniques to Analyse Sleep EEG Data

“…hybrid systems [8], interval histograms [9], wavelet analy sis [10]. and period analysis [11,12], Many authors used power spectrum variables for analysis [4].…”

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

“…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.…”