Classification of EEG Signals Recorded While Imagining Odors with Effective Band Pass Filter Parameters

Abstract

Technological advances in medicine allow us to learn more about human physiology, anatomy, endocrinology and neurology. In this way, many diseases can be diagnosed early and treatment processes can be closely monitored. On the one hand, the advancing technology allows us to demonstrate how the human brain reacts to various environmental stimuli. In this study, electroencephalography (EEG) technique was used to record the reactions of the brain to the imagination of cloves and mint odors. The distinguishability of brain responses to these odors was tested using pixel means of gray-level images obtained from Burg-based color spectrogram images. By using pixel means as an attribute, a classification accuracy of 82.5% was obtained with k-nearest neighbor. The results showed that the EEG signals that can occur when the brain imagines different odors can be distinguished.