Nöro-elektriksel sinyaller için yeni bir zaman-frekans analiz tekniği

ABSTRACT A NEW TIME-FREQUENCY ANALYSIS TECHNIQUE FOR NEUROELECTRIC SIGNALS D. İlhan Tüfekçi M.S. in Electrical and Electronics Engineering Supervisor: Assoc. Prof. Dr. Orhan Arıkan August 2002 In the presence of external stimuli, the functioning brain emits neuroelectrical signals which can be recorded as the Event Related Potential (ERP) signals. To understand the brain cognitive functions, ERP signals have been the sub ject matter of many applications in the field of cognitive psychophysiology. Due to the non-stationary nature of the ERP signals, commonly used time or frequency analysis techniques fail to capture the time-frequency domain localized nature of the ERP signal components. In this study, the newly de veloped Time-Frequency Component Analyzer (TFCA) approach is adapted to the ERP signal analysis. The results obtained on the actual ERP signals show that the TFCA does not have a precedent in resolution and extraction of uncontaminated individual ERP signal components. Furthermore, unlike the existing ERP analysis techniques, the TFCA based analysis technique can reliably measures the subject dependent variations in the ERP signals, which mopens up new possibilities in the clinical studies. Thus, TFCA serves as an ideal tool for studying the intricate machinery of the human brain. Keywords: Event Related Potentials, ERP, Oscillatory Brain Signals, EEG, Brain Signal Analysis, Time-Frequency Signal Analysis, Event Related Oscil lations, ERO IV