EEG nonstationarity during intracranially recorded seizures: statistical and dynamical analysis

Area of Science:

• Neuroscience
• Signal Processing
• Computational Biology

Background:

• Investigating complex signals like electroencephalographic (EEG) recordings requires advanced analytical techniques.
• Epileptic seizures exhibit nonstationarity, necessitating novel approaches for detailed analysis.

Purpose of the Study:

• To divide EEG recordings during epileptic seizures into sequential stages using spectral and statistical analysis.
• To reconstruct discrete-time models reflecting the deterministic properties of EEG voltage time series.

Main Methods:

• Analysis of intracranial human EEG recordings from three epilepsy patients.
• Application of statistical methods (power spectra, wavelet spectra, probability distribution functions).
• Utilized dynamical nonstationarity analysis and comparison of dynamical models.

Main Results:

• Dynamical nonstationarity analysis provided finer detail within seizures compared to statistical methods.
• Typically identified three to four distinct dynamic stages within seizures.
• Dynamical analysis enhanced the distinction between interictal activity and seizure events.

Conclusions:

• Nonstationarity analysis unveils temporal structures within epileptic seizures, aiding understanding of seizure evolution.
• This method shows potential for accurate seizure onset identification.
• The approach offers novel insights into seizure temporal dynamics, surpassing conventional methods.