Detection and removal of stimulation artifacts in electroencephalogram recordings.

Stimulation artifacts are short-duration, high-amplitude spikes which can be observed in electroencephalogram (EEG) recordings whenever surface functional electrical stimulation (FES) is applied during recordings. Stimulation artifacts are of non-physiologic origin and hence have to be removed before analysis of the EEG can take place. In this paper, algorithms for the detection and removal of stimulation artifacts are presented. The algorithms require only little computational resources and can be applied online, while signals are recorded. Therefore, the algorithms are suitable for applications such as online control of FES based neuroprostheses by a brain-computer interface. Tests are performed with datasets recorded from two subjects for artifact durations ranging from 0.5 ms to 10 ms. After application of the artifact removal algorithms the signal-to-noise ratio of the reconstructed signals ranges from 15 dB to 45 dB, depending on the duration of artifacts and the type of algorithm.