Automatic interpretation of the adult waking EEG

Abstract

Automatic interpretation of EEG has been faced with significant difficulties because of continuous fluctuation of the background activity, occurrence of paroxysmal activities, contamination with artifacts, and use of different electrode montages. Therefore, the previous attempts of automatic EEG interpretation have focused on a certain feature of EEG such as the background activity and paroxysmal abnormalities. The primary purpose of this study is to establish a computer-assisted, off-line system for automatic comprehensive interpretation of EEG, which takes into account all features of the adult waking EEG and provides the results in a written form. First, we extracted features of EEGer’s visual inspection to quantitatively score the background activity, and the parameters for each frequency band were selected. Automatic detection of artifacts and spikes and automatic judgment of vigilance and attention level of subjects were incorporated into the system. The results of this system were shown to be in good conformity with the results of visual inspection of the same records by EEGers in healthy subjects as well as in patients with brain disorders including epilepsy. This system can be used as a supplementary tool for EEGer’s visual inspection, and its modified form can be applied for training EEGers and EEG technicians.