The human EEG — Physiological and clinical studies

Abstract

The present review summarizes the research in EEG performed by our group during the last 5 years. Our studies have been focussed on two areas: studies of variability and correlations in the oscillations during resting conditions of normal subjects, and the abnormalities related to type 1 diabetes. Recordings in normal subjects showed that also under standardized conditions with regular cycles of closed and open eyes, there is a temporal variability of the spectral components in EEG that necessitates samples > 124 s in order to achieve estimates of alpha power with a coefficient of variation < 0.1 in all recording channels (brain regions). The temporal variability in alpha and beta power demonstrates long-range temporal correlations, i.e., periods of large power (alpha or beta) are more likely to be followed by large power, and vice versa. The long-range temporal correlations were reproducible, especially during the closed-eyes condition, stronger in males than females, and not age dependent.

In patients with type 1 diabetes, the alpha and beta power components were both decreased with similar nadirs in the posterior temporal regions, and the slow spectral components were increased in the frontal regions. The cognitive function was presently not studied but in a group of adolescents with diabetes we found a correlation between the presence of slow activity and the number of hypoglycaemic episodes. The loss of alpha and beta power was highly correlated which initiated a study of the normal alpha–beta correlation. A significant 1:2 phase synchronization was present between alpha and beta oscillations with a phase lag of about π/2 in all electrode derivations. The strong frequency relationship between the resting beta and alpha oscillations suggested that they are generated by a common mechanism.

Introduction

During the past 5 years the focus of my research has been on the human EEG and its application in studies on patients with diabetes mellitus. This originated in my interest since many years of diabetes complications in the peripheral nervous system [1], [2], [3], [4], [5], [6], [7]]. Cognitive tests and EEG have shown an increased incidence of abnormalities in patients with type 1 diabetes (insulin dependent). Although it is well-established that a tight blood sugar control is beneficial in order to prevent complications of nephropathy, retinopathy and peripheral neuropathy in patients with type 1 diabetes, there is a concern that this may carry a risk for negative effects on the brain, especially in relation to episodes of very low blood glucose.

Electroencephalographic signals demonstrate a large variability in time, which raises the question of the amount of data that is required for an accurate description of the spectral pattern in EEG. This motivated us to perform a study on the variability of different EEG spectral parameters, considering the time course of the EEG spectrum in resting conditions, and the relationship between the spectral parameters and the length of the analyzed segments, in order to get a methodological basis for clinical studies [8]. The major conclusion of this study was that the power estimates of the resting EEG activity are heavily dependent on the length of the analyzed segments, and the way these segments are selected. This observation is particularly relevant for clinical and drug studies where short data segments often are selected from a recording, which may bias the estimation of the EEG parameters.

Recent studies indicate that the amplitude of EEG oscillations in the human brain possesses long-range temporal correlations which indicates that events in the past affect the development of the process in the future [9]. Long-range temporal correlations are thought to be advantageous for a reliable transfer of information in neuronal populations [10]. We have studied the test–retest reliability of the long-range temporal correlations in the spontaneous neuronal oscillations in the open- and closed-eyes conditions representing different levels of arousal [11]. The amplitude fluctuations of alpha and beta oscillations were used since these two rhythms are reactive to changes in arousal and usually have well defined peaks in the spectrum [12]. Furthermore, if the long-range temporal correlations are susceptible to changes in the internal state of a subject then their use can be advocated in different cognitive paradigms as an approach to study complex dynamic behaviour of neuronal systems.

We have also studied the topography in the long-range temporal correlations in the amplitude fluctuations of alpha and beta neuronal oscillations and a possible dependency on age and gender of the subjects [13]. A topographic difference in the long-range temporal correlations would imply that these correlations might be affected by specific mechanisms related to the generation of a given neuronal process. Both age and gender are known to affect average amplitude of EEG oscillations [14].

A decline in the cognitive function is a complication of diabetes mellitus that has received relatively little attention in comparison with other neurological complications of diabetes. The interest for this has increased since there is concern that repeated episodes of severe hypoglycaemia may damage the brain, and that this might become more common with the introduction of a more intensive insulin therapy [15]. Several studies have demonstrated a decline in cognitive function in type 1 diabetes which was correlated to a background of frequent hypoglycaemic attacks especially in children [16]. However, other studies have demonstrated a decline in cognitive function that was related to disease duration, age of disease onset, male gender and metabolic status at time of diagnosis — but unrelated to hypoglycaemic episodes [17]. Some EEG studies over the years have shown an increase in slow activity and a decreased alpha frequency in both children and adults with type 1 diabetes [18], [19], [20].

We have performed two studies on the effect of type 1 diabetes on the EEG spectral parameters, one in young adults [21] and the other in adolescents [22]. The recording conditions were standardized and the entire 15 min recording was included for the quantitative analysis in order to achieve a high sensitivity [8]. A striking and consistent finding in these studies was that both alpha and beta powers were decreased, with similar topography and with a 1:2 frequency relation between the peak frequencies of the alpha and beta band.

Because of the often non-sinusoidal structure of neuronal oscillations, changes in the amplitude of the alpha rhythm might be accompanied by similar changes at higher frequencies, often in the frequency range of the harmonics of the alpha rhythm [23]. However, little attention has been drawn to this issue and often high frequency processes in EEG are not discussed in the context of changes in the alpha rhythm. We have quantified the amplitude, phase and frequency relationship of alpha and beta oscillations in normal subjects in a resting condition using wavelet analysis [24]. High frequency components > 15 Hz have been correlated with cognitive functions [25], motor performance [26] and perception [27]. However, without knowledge on whether the fast frequency components of an EEG originate from alterations in the shape of the alpha rhythm or not, it is not possible to conclude that they have specific neuronal mechanisms separating them from alpha oscillations.