Neuronal population oscillations at a variety of frequencies can be readily seen in electroencephalographic (EEG) as well as local field potential recordings in many different species. Although these brain rhythms have been studied for many years, the methods for identifying discrete oscillatory epochs are still widely variable across studies. The "better oscillation detection" (BOSC) method applies standardized criteria to detect runs of "true" oscillatory activity and rejects transient events that do not reflect actual rhythms. It does so by estimating the background spectrum of the actual signal to derive detection criteria that include both power and duration thresholds. This method has not yet been applied to nonhuman data. Here, we test the BOSC method on two important rat hippocampal oscillatory signals, the theta rhythm and slow oscillation (SO), two large amplitude and mutually exclusive states. The BOSC method detected both the relatively sustained theta rhythm and the relatively transient SO apparent under urethane anesthesia and was relatively resilient to spectral features that changed across states, complementing previous findings for human EEG. Detection of oscillatory activity using the BOSC method (but not more traditional Fourier transform-based power analysis) corresponded well with human expert ratings. Moreover, for near-continuous theta, BOSC proved useful for detecting discrete disruptions that were associated with sudden and large amplitude phase shifts of the ongoing rhythm. Thus, the BOSC method accurately extracts oscillatory and nonoscillatory episodes from field potential recordings and produces systematic, objective, and consistent results-not only across frequencies, brain regions, tasks, and waking states, as shown previously, but also across species and for both sustained and transient rhythms. Thus, the BOSC method will facilitate more direct comparisons of oscillatory brain activity across all types of experimental paradigms.
Copyright © 2011 Wiley Periodicals, Inc.