Home
Rhythmic patterns in neural time series provide a window on the neural dynamics that shape perception, cognition and action, and therefore are a major signal of interest in the cognitive, computational and systems neurosciences. However, typical descriptions of rhythmicity lack detail, e.g., failing to indicate when and for how long rhythms occur. Such lack of detail becomes problematic in the face of non-stationarities in rhythmic engagement, i.e., when the presence of rhythms varies over time. To overcome resulting pitfalls, there is a need for methods that identify the occurence of non-stationary rhythmic periods in a systematic manner.
eBOSC (extended Better OSCillation detection) is a toolbox (or a set of scripts) that can be used to detect the occurrence of rhythms in continuous signals (i.e., at the single trial level). It uses a static aperiodic ‘background’ spectrum as the basis to define a ‘power threshold’ that continuous signals have to exceed in order to qualify as ‘rhythmic’. As such, it leverages the observation that stochastic components of the frequency spectrum of neural data are characterized by a '1/f'-like power spectrum. An additional ‘duration threshold’ can be set up in advance, or rhythmic episodes can be filtered by duration following detection to ensure that detected rhythmic episodes have a rather sustained vs. transient appearance. The main goal of the code is to produce a list of rhythmic episodes.
The identification of rhythmic episodes provides a tool to subsequently investigate the characteristics of rhythmic periods, e.g. w.r.t. their timing, duration and magnitude.
The above image shows some example applications. Left: By separating alpha-frequency episodes > 3 cycles (top) and < 3 cycles (bottom), eBOSC can separate more sustained from transient spectral events in empirical data. Right: By specifically rhythmic episodes, rhythmic spectra can be 'boosted' due to removal of aperiodic contributions in time. Figure adapted with permission from Kosciessa et al. (2020).
You can find an introductory tutorial here!
The code base is always subject to change, but older versions can be accessed by checking out previous commits. If you have been using a previous version, major changes relating to I/O are tracked in the legacy information.
The principles are described in more detail in Kosciessa et al. (2020), where the eBOSC code is also benchmarked.
Kosciessa, J. Q., Grandy, T. H., Garrett, D. D., & Werkle-Bergner, M. (2020). Single-trial characterization of neural rhythms: Potential and challenges. NeuroImage, 206, 116331. doi:10.1016/j.neuroimage.2019.116331
The eBOSC code extends the codebase of BOSC, developed by Caplan et al. (see below).
Whitten, T. A., Hughes, A. M., Dickson, C. T., & Caplan, J. B. (2011). A better oscillation detection method robustly extracts EEG rhythms across brain state changes: The human alpha rhythm as a test case. NeuroImage, 54(2), 860–874. http://doi.org/10.1016/j.neuroimage.2010.08.064