A comparison between healthy and neurological disorders patients using nonlinear dynamic tools

In this paper, we evaluate the differences between Healthy subjects (HS), Parkinson Disease (PDS) subjects, and Epileptic subjects (EBS — Epileptic subject between seizures and ES — epileptic subjects in seizure) by computing some linear, statistic and non-linear features, such as: correlation dimension, maximum Lyapunov exponent and Hurst coefficient. A comparison is made between PDS, EBS, ES and HS groups using non-linear parameters. The EEG data are wavelet multiresolution decomposed into subbands of interest (alpha 8–12 Hz, beta 13–30 Hz, delta 0–4 Hz, theta 4–8 Hz, gamma 30–60 Hz). We applied rescaled range method to estimate the Hurst coefficient of the decomposed signals, with different types of wavelet transforms. The results obtained using the maximum Lyapunov exponent do not highlight very well differences between the 4 groups analyzed (HS, PDS, EBS and ES) but using Hurst coefficient we can make a very good differentiation of EEG signals. The results show that the Hurst coefficient is significant different for delta and theta rhythms extracted from the EEG signal in case of the healthy subjects compared with PDS or EBS subjects. The Hurst coefficient for healthy subjects has a value higher than 0.5 and for PD subjects or epileptic subject between seizures has a value lower than 0.5. Hence, the non-linear parameters such as Hurst coefficient can help in EEG interpretation and in neurological disorders diagnosis.