Time-resolved parameterization of aperiodic and periodic brain activity.

Area of Science:

• Neuroscience
• Computational Neuroscience
• Signal Processing

Background:

• Neural signals contain both periodic and aperiodic components.
• Existing methods for spectral decomposition are often static, failing to capture dynamic neural changes.
• Understanding time-varying neural dynamics is crucial for linking brain activity to behavior.

Purpose of the Study:

• Introduce Spectral Parameterization Resolved in Time (SPRiNT), a novel method for time-resolved neural power spectrum decomposition.
• Evaluate SPRiNT's ability to accurately capture time-varying spectral features.
• Investigate the temporal dynamics of aperiodic neural activity and its relationship to behavior.

Main Methods:

• Developed SPRiNT for decomposing neural power spectra into time-varying periodic and aperiodic components.
• Validated SPRiNT using naturalistic synthetic data, comparing it with wavelet-based methods.
• Applied SPRiNT to resting-state EEG data (n=178) and rodent intracranial recordings.

Main Results:

• SPRiNT reliably recovers time-varying spectral features in synthetic data.
• Demonstrated temporal fluctuations in aperiodic spectral parameters in human EEG data.
• Showed correlations between aperiodic neural dynamics and movement behavior in rodent models.

Conclusions:

• SPRiNT provides a robust method for analyzing non-stationary neural signals.
• The method advances the quantification of complex neural dynamics at behaviorally relevant timescales.
• SPRiNT is poised to support research on time-varying neural power spectra.