[Computer analysis of electroencephalogram].

Abstract

Many computing systems have been developed and applied to the sleep studies. The most famous system is EEG sleep stage determination system. It becomes possible to process the PSG data during one night in a few minutes by these systems. We also have been studying for such system and analyzing method of EEG. This paper presents the feature extraction and recognition methods of EEG waves. For the feature extraction methods, wave shape analysis, wavelet transformation, AR (autoregressive) model and damped system with Poisson impulse input sequences are used. For the recognition methods, statistical pattern recognition method and artificial neural networks are used. The effectiveness of these methods is confirmed through the experimental studies.