Computational modelling of electroencephalogram in Alzheimers' disease

Abstract

The aim of this study is to examine the brain dynamics in patients with Alzheimer````s disease by the simulation of the EEG in the progress of Alzheimer````s disease using neural network constructed based on biological anatomic structure of cerebral cortex, the mechanism of EEG generation and the Freeman````s asymmetry sigmoid nonlinearity. The simulated EEG have the comparable dynamical properties in accord with physiological findings of the human EEG. The spectrum of the simulated EEG is similar to that of the real EEG seen in the alert state with components in the delta, theta, alpha, beta ranges. The simulated EEG has the similar values of the correlation dimension (D2), the first positive Lyapunov exponent (L1), and Kolmogorov-Sinai entropy (KS-entropy), which are measures of complexity. These nonlinear estimates revealed that the simulated EEG is a complex and chaotic time series. Determinism test showed that it does not have low-dimensional determinism but nonlinearity as real EEG does. We investigate the changes of linear and nonlinear properties of the simulated EEG as the synaptic densities of the neural network decrease like the cerebral cortex of the brain damaged by Alzheimer````s disease. We show that the mean frequency of the power spectrum of the simulated EEG shifts to low frequencies as the synaptic densities decrease. The values of the D2, the L1, and the KS-entropy also decrease as the network deteriorates. These results are comparable nature of the EEG in patients with Alzheimer````s disease. Especially the decreases of synaptic densities of long corticocortical connection (alpha connection) with fixed subcortical connections (beta connections) lead to the power spectrum of the simulated EEG shift to low frequencies and lowered complexities estimated by the D2, the L1 and the KS entropy, while the decreases of subcortical connections with fixed long corticocortical connections result in no significant changes in those nonlinear measures. Th...