Application of neural networks for EEG analysis. Considerations and first results

Abstract

This paper presents the results of the practical use of artificial neural networks in the field of EEG analysis. It describes the general methodology of application as well as a case study of a discrimination of depressive and psychotic patients using 16-channel long-term EEG data prepared by classical pre-processing (spectral decomposition). This study shows advantages and current limits concerning different levels of generalisation capabilities using a representative application example.