Slope of the power spectral density flattens at low frequencies (<150 Hz) with healthy aging but also steepens at higher frequency (>200 Hz) in human electroencephalogram

Abstract

AbstractBrain signals such as electroencephalogram (EEG) often show oscillations at various frequencies, which are represented as distinct “bumps” in the power spectral density (PSD) of these signals. In addition, the PSD also shows a distinct reduction in power with increasing frequency, which pertains to aperiodic activity and is often termed as the “1/f” component. While a change in periodic activity in brain signals with healthy aging and mental disorders has been reported, recent studies have shown a reduction in the slope of the aperiodic activity with these factors as well. However, these studies only analysed PSD slopes over a limited frequency range (<100 Hz). To test whether the PSD slope is affected over a wider frequency range with aging and mental disorder, we collected EEG data with high sampling rate (2500 Hz) from a large population of elderly subjects (>49 years) who were healthy (N=217) or had mild cognitive impairment (MCI; N=11) or Alzheimer’s Disease (AD; N=5), and analysed the PSD slope till 800 Hz. Consistent with previous studies, the 1/f slope up to ~150 Hz reduced with healthy aging. Surprisingly, we found the opposite at higher frequencies (>200 Hz): the slope increased with age. This result was observed in all electrodes, for both eyes open and eyes closed conditions, and for different reference schemes. Slopes were not significantly different in MCI/AD subjects compared to age and gender matched healthy controls. Overall, our results constrain the biophysical mechanisms that are reflected in the PSD slopes in healthy and pathological aging.Significance StatementAperiodic activity in the brain is characterized by measuring the slope of the power spectrum of brain signals. This slope has been shown to flatten with healthy aging, suggesting an increase in some sort of “neural noise”. However, this flattening has been observed only over a limited frequency range (<150 Hz). We found that at higher frequencies (>200 Hz), the opposite happens: the slope steepens with age. This occurs at all electrodes, irrespective of state and referencing techniques. However, the slope is unchanged in subjects with early Alzheimer’s Disease (AD) and their controls. Our results shed new light on the properties of neural noise and the neurophysiological processes affecting AD and the aperiodic activity.