Robust estimation of 1/f activity improves oscillatory burst detection.

Area of Science:

• Neuroscience
• Computational Neuroscience
• Signal Processing

Background:

• Neural oscillations manifest as transient bursts with dynamic amplitude and frequency.
• Accurate quantification of these bursts is crucial for understanding brain-behavior relationships, particularly in continuous electrophysiological datasets.
• Distinguishing rhythmic oscillatory activity from arrhythmic background 1/f activity is a key challenge in analyzing neural recordings.

Purpose of the Study:

• To introduce and validate fBOSC, a modified framework for robustly detecting neural oscillatory bursts.
• To improve the modeling of background 1/f activity in neural power spectra, addressing limitations of previous methods like BOSC.
• To enhance the sensitivity and standardization of oscillatory burst detection across different frequency bands.

Main Methods:

• Developed fBOSC, a MATLAB toolbox utilizing spectral parametrization to model background 1/f activity.
• Evaluated fBOSC's performance through simulations, comparing it against existing methods for 1/f power spectrum modeling.
• Applied fBOSC to resting-state magnetoencephalography (MEG) and intracranial electrophysiology (iEEG) datasets for theta-band oscillatory burst detection.

Main Results:

• fBOSC demonstrated superior accuracy in modeling the 1/f power spectrum, especially in spectra with a 'knee' feature common in neural data.
• Unlike other methods, fBOSC remained unaffected by oscillatory peaks within the power spectrum.
• fBOSC enabled standardized sensitivity for detecting oscillatory bursts across frequencies (e.g., theta and alpha bands).

Conclusions:

• fBOSC provides a more accurate and robust method for modeling background 1/f activity in neural data.
• This improved background modeling enhances the reliable detection of neural oscillatory bursts, even in long, continuous datasets.
• The fBOSC framework offers a valuable tool for advancing research into brain function and behavior through precise analysis of neural oscillations.