Nonlinear Dynamical Analysis of EEG for Early Detection of Degenerative Diseases in the Brain

Alzheimer's disease (AD) is the most common degenerative brain disease characterized by mental deficits and behaviour disturbance. The electroencephalogram (EEG) has been for a long time a useful tool for diagnosis and prognosis of degenerative brain diseases. EEGs of patients with AD exhibit some abnormalities characterized by a less complex activity associated with functional impairments among cortical areas assumed to be due to death of cortical neurons among others. The EEG is assumed to be a nonlinear system. Nonlinear methods derived from the chaos theory are used to characterise nonlinear EEG activity in patients with Alzheimer's disease. The correlation dimension (D2) and the largest Lyapunov exponent (L1) are two common nonlinear methods which reflect the complexity of the cortical dynamics underlying EEG signals. The applicability of nonlinear methods is first tested. The presence of nonlinear dynamics is determined by comparing the time reversal asymmetry statistics of the original EEG data set with the one of the phase-randomized surrogate EEG data set constructed with the original data. The results show that EEG signals have been created by a nonlinear process. The EEG signals from patients with AD (in the early and late stage of the disease) and normal controls were recorded at 21 locations using the 10-20 International System. D2 and L1 are then computed to estimate the complexity of EEG patterns in patients with AD and healthy controls. D2 and L 1 values from patients with AD are for most electrodes lower than those from healthy controls which indicate a decrease complexity of the dynamics underlying diseased brains. The D2 and L1 values of patients with early AD suggest that the occipital and temporal regions of the brain are the most affected by the disease and that the disease is more apparent in the left hemisphere of the brain. The brains of patients with AD thus show behaviours less chaotic than the one of normal controls. The different results imply that nonlinear dynamical analysis (NDA) of EEG may be a useful tool for early diagnosis of AD. Keywords: Alzheimer's disease, Chaos, EEG analysis, Correlation Dimension, Lyuapunov ExponentFaculty of Technolog