The objective of this project is to detect the onset of drowsiness/sleep by an analysis of the characteristic features using EEG and MEG waveforms corresponding to the “Awake” and “Drowsy” states using MCT (Mean Comparison Test) approach. Drowsiness refers to an abnormally sleepy feeling, especially at a time when the subject is required to be awake and alert. Such momentary feelings of drowsiness may lead to catastrophic errors of judgment. The proposed system detects the condition of drowsiness by acquiring the physiological signals such as EEG and MEG; the time and frequency domain characteristics of those signals change depending upon the level of alertness. The acquired EEG and MEG signals are generally distorted due to the addition of noise from various sources as well as by parasitic biological artifacts. So, filtering plays a vital role in the preprocessing of the data. The useful features giving information on awake, drowsy and sleep states of the subject can be extracted by implementing wavelet decomposition and windowing technique. These extracted features are then supplied as inputs to the classifier, which classifies them into awake and drowsy state by following a set of rules and procedures assigned initially.

Keywords: EEG, MEG, MCT, ROC