Classification of Electroencephalogram signals using Artificial Neural Networks

Abstract

The study of Artificial Neural Networks (ANN) has proved to be fascinating over the years and the development of these networks has grown strongly in recent years. The neural networks have come to be increasingly convincing methods for solving complex problems, through artificial intelligence. In particular this work focused on development of an artificial neural network for identifying diseases: Parkinson's, Huntington's and Amyotrophic Lateral Sclerosis, based on signals from the Electroencephalogram (EEG). The phases of the project were developed through a number of operations implemented in Matlab. The Fourier transform was seen as the main technique of signal processing, in order to analyze and diagnose diseases in the study. The work consisted in the first stage process the EEG signals to serve as an entry into the ANN in order to reveal a distinctive feature in the different diseases studied, and then, create a model capable to distinguish the diseases. For this purpose 4 methodologies were used with different processing of the EEG signal. The 4 methodologies are compared in this paper.