Artificial Eeg Signal Generated By A Network Of Neurons With One And Two Dendrites

The electroencephalogram EEG signal analysis is being strongly explored as a new potential tool for control, communication, and clinical diagnosis applications related to many neurological pathologies such as epilepsy, autism, Alzheimer. Analyzing EEG data is a very interesting approach to study cognitive processes. Thus, it can help researchers to understand the brain processes, doctor to establish an efficient medical diagnosis. To better understand the nature and the behavior of this signal which seems to be complex, nonlinear, non-stationary, and imbalanced, we propose to generate this signal using a predefined model of neurons with one and two dendrites by studying their activation functions. Then, we build a network of neurons with two dendrites correlated with the alpha and beta frequencies which represent the most significant sub-bands in EEG signals to generate a signal quite close to EEG recordings. Through this paper, we aim to present a comprehensive study and an exploration of artificial EEG signal generation using only neurons with two dendrites that provided different activation and excitation functions. The presented result shows that the generated EEG signal with six neurons of two dendrites is quite similar to the real EEG recording for a healthy subjects. In addition, some artifacts like Alzheimer and Parkinson have been emulated and identified among the generated spikes. Finally we proved that we can eliminate some artifacts related to harmonics by applying activation function with opposite polarities.