Comparison of human ictal, interictal and normal non-linear component analyses

Area of Science:

• Neuroscience
• Dynamical Systems Theory
• Signal Processing

Background:

• Electroencephalogram (EEG) signals exhibit complex, non-linear dynamics.
• Understanding the underlying mechanisms of EEG generation is crucial for neurological diagnostics.
• Epilepsy, particularly complex partial seizures, presents unique EEG patterns.

Purpose of the Study:

• To investigate if EEG in different neurological states arises from integrated non-linear dynamic systems.
• To analyze the non-linear properties of EEG and its filtered rhythms in healthy subjects and epileptic patients.
• To explore the potential of differentiating neuronal network characteristics through EEG component dynamics.

Main Methods:

• Digitally filtered EEG components (delta, theta, alpha, beta, gamma) from control subjects and epilepsy patients.
• Calculated correlation dimension on original EEG signals and surrogate data.
• Developed an accelerated method for correlation integral calculation and visualized correlation dimension meaning.

Main Results:

• EEG signals showed significantly lower correlation dimensions than surrogate data in both controls and patients.
• Filtered EEG components displayed varied complexity; delta, alpha, beta, and gamma in controls resembled surrogate data.
• Alpha component in interictal EEG matched surrogate data complexity; theta and alpha dimensions were stable between interictal and ictal states.

Conclusions:

• EEG and its filtered components exhibit diverse dynamic properties across neurological states.
• The dynamics of filtered EEG components can differentiate neuronal network characteristics.
• Decomposing EEG into distinct dynamic systems aids in understanding human EEG generation mechanisms.