Classification of tone stimulated EEG signals using independent components and power spectrum vectors

• Main Authors: Navea, Roy Francis R., Dadios, Elmer Jose P.
• Format: text
• Published: Animo Repository 2016
• Online Access: https://animorepository.dlsu.edu.ph/faculty_research/1059; https://animorepository.dlsu.edu.ph/context/faculty_research/article/2058/type/native/viewcontent
• Institution: De La Salle University

© 2015 IEEE. The brain responds to different stimuli. In this study, the brain was stimulated by an audio sound that plays the tones C, F and G of the piano keyboard using a predefined audio piece. The brain's response was recorded and a classification scheme was proposed. The EEG information was segmented into baseline, C, F, G and s-baseline. The independent components and the power spectrum vectors of each segment were obtained. The independent components and power spectrum vectors of the baseline (when relaxed) is highly distinguishable as compared to the other segments. The other segments are more scattered in the frequency domain more than in the statistical domain. Artificial neural networks (ANN) were used to classify the segments using a leave-out-one cross validation method. Both features are useful and gave high classification percentages. However, higher classification percentages were obtained using the power spectrum vectors.