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

Abstract

Objectives: The non-linear properties of EEG and filtered rhythms obtained from healthy subjects and epileptic patients with complex partial seizures were analyzed to investigate whether EEG in different neurological states can be generated by the mechanism that integrates several non-linear dynamic systems. Methods: The control EEG (from 26 healthy subjects), interictal EEG and ictal EEG (from 25 patients) were digitally filtered into delta (0.5–4 Hz), theta (4–8 Hz), alpha (8–13 Hz), beta (13–30 Hz) and gamma (30–40 Hz) components. The correlation dimension was calculated on each original signal and corresponding surrogate data. A new method was developed to accelerate the calculation of the correlation integral. Function P( m, r) was defined to visualize the meaning of the correlation dimension. The point critical to the estimation was determined by the P( m, r) function. Results: The EEG in the control subjects and patients showed significantly lower correlation dimensions than the surrogate data. The delta, alpha, beta and gamma components from the control EEG exhibited similar complexity to the surrogate data, while only the alpha component from the interictal EEG presented the same dimension as the surrogate data. The correlation dimensions of the theta and alpha components remained the same when the neurological state changed from interictal EEG to ictal EEG. The complexity of the beta component was higher than the complexity of other components in both control subjects and patients. The correlation dimension of EEG was significantly correlated to the complexity of delta, theta, beta and gamma components. Conclusions: Our results suggest that EEG and filtered components in different neurological states demonstrate varied dynamic properties. The characteristics of neuronal networks can be differentiated by the dynamics of filtered components. Separating EEG into different dynamic systems may facilitate understanding of the mechanisms involved in the human EEG.