Evaluating and Comparing Measures of Aperiodic Neural Activity

Area of Science:

• Neuroscience
• Signal Processing
• Computational Biology

Background:

• Aperiodic neural activity, also known as 1/f or fractal activity, is a significant feature in neuro-electrophysiological recordings.
• This activity dynamically relates to aging, clinical diagnoses, conscious states, and behavior.
• Current analysis methods for aperiodic activity lack clear interrelationships, hindering synthesis and understanding.

Purpose of the Study:

• To systematically survey and compare diverse methods for measuring aperiodic neural activity.
• To clarify the relationships between different analytical approaches for neural aperiodic signals.
• To provide a unified framework for interpreting aperiodic neural activity.

Main Methods:

• Automated literature analysis to identify common aperiodic activity measurement methods.
• Statistical time series simulations to evaluate and compare selected methods.
• Application of methods to empirical electroencephalography (EEG) and intracranial EEG (iEEG) datasets.

Main Results:

• Established consistent relationships between various aperiodic measures, revealing shared variance.
• Demonstrated that frequency domain methods specifically capture aperiodic features.
• Showed that time domain measures are more susceptible to oscillatory activity.
• Validated simulation findings with real EEG and iEEG data.

Conclusions:

• Much of the variance captured by different aperiodic measures is shared, but idiosyncrasies exist.
• Frequency domain analyses offer greater specificity for aperiodic neural activity.
• Understanding these measure relationships facilitates reevaluation of past findings and guides future research.
• Provides a foundation for more robust analysis of neural aperiodic signals.