Simulated power spectral density (PSD) of background electrocorticogram (ECoG)

Area of Science:

• Neuroscience
• Computational Neuroscience
• Signal Processing

Background:

• Electrocorticography (ECoG) background activity in the cerebral cortex during rest and slow-wave sleep often resembles broadband noise.
• Power spectral density (PSD) analysis frequently reveals a power-law distribution (1/f(x)) with exponents ranging from 2 to 4.

Purpose of the Study:

• To explain the neural source of background ECoG activity using a computational model.
• To investigate the characteristics of ECoG during rest, sleep, and active states.
• To establish the resting ECoG state as an optimal baseline for analyzing neural structures.

Main Methods:

• Developed a model simulating neural sources based on mutual excitation among pyramidal cells.
• Modeled dendritic responses to impulse inputs as a sum of two exponential terms.
• Convolved the neural response kernel with a Poisson process to generate simulated ECoG time series and PSD.

Main Results:

• The model successfully replicated ECoG time series and PSD characteristics, fitting "brown" or "black" noise patterns.
• The PSD slope was found to be dependent on the rate of rise in the neural response.
• Deviations from rest/sleep to active states introduced local peaks in the PSD, indicating emergent nonrandom structure.

Conclusions:

• Resting ECoG activity is characterized as low-dimensional noise, explained by a model of interacting excitatory neurons.
• The observed variations in PSD slope are attributed to homeostatically regulated background activity levels.
• The resting state provides an optimal baseline for identifying and quantifying artifactual and physiological structures in activated ECoG signals.