Artifact Detection and Correction for Operator Functional State Estimation

Determining operator functional state is a critical component of adaptively aiding closed human-in-the-loop systems. The input measures or features used to define the operator functional state (OFS) model must be free of artifacts to ensure accurate classification of operator functional state. Electroencephalography (EEG) is a major component of the OFS model. Since the EEG magnitude is small relative to other psychophysiological measures it is most susceptible to contamination or artifact. Eye blinks and body movements are the usual culprit of artifact in EEG. This paper discusses an adaptive filter algorithm for removal of both horizontal and vertical eye movements including blinks. Muscle activation is also a major source of EEG contamination. A discussion of a promising muscle artifact removal using independent component analysis is presented. Additional artifacts can be detected but not removed using a simple statistical technique which is presented here.