Classifying eye and head movement artifacts in EEG signals

Abstract

Brain Computer Interfaces has some exciting prospects such as controlling devices at the speed of thought. However BCI technology is far from attaining this goal. A significant challenge the EEG-based system has is the interference of artifacts in the EEG generated by eye and head movement. This paper presents the use of machine learning techniques to classify artifacts in the EEG. Successful artifact classification was then be applied to improve existing artifact removal techniques. The experiment used a state-of-the-art EEG system to gather the classifier input. An eye tracker and motion sensor were also used to measure and provide the ground truth for the classification experiments. The data from these devices were captured using custom built software developed for this research. The classifiers tested showed potential to classify artifacts in the EEG when trained on a per-person basis. This research paves the way for further work to be carried out to explore subject-independent artifact classification.