Single-trial processing of event-related potentials using outlier information

Abstract
A new approach to extracting single-trial event-related information is described in this paper. This approach, called the outlier processing method (OPM), is based on the concept that event-related information is contained in EEG time-series outliers. In particular, the OPM has been effective in extracting motor-related information from single-trial EEG. An investigation into the viability of the OPM was carried out on single-trial EEG data from four subjects. The EEG was collected under two conditions: an active task in which the subject performed a skilled thumb movement and an idle task in which the subject remained alert but did not carry out any motor activity. The results of this investigation demonstrated that consistent single-trial motor related information can be successfully extracted using the OPM.