Spectral waveform analysis dissociates human cortical alpha rhythms

Abstract

The non-sinusoidal waveform of neuronal oscillations reflects the physiological properties of underlying circuit interactions and may serve as an informative biomarker of healthy and diseased human brain function. However, little is known about which brain rhythms can be dissociated based on their waveform and methods to comprehensively characterize waveforms are missing. Here, we introduce a novel spectral waveform analysis (SWA) that provides a complete waveform description, is noise-resistant, and allows to reconstruct time-domain waveforms. We applied this framework to human magnetoencephalography (MEG) recordings during rest and identified several distinct and previously unknown cortical alpha waveforms that were temporally stable and specific for individual subjects. Our findings suggest at least four distinct alpha rhythms in human sensorimotor, occipital, temporal, and parietal cortex. SWA provides a powerful new framework to characterize the waveform of neural oscillations in the healthy and diseased human brain.