Fully Automated Artifact Removal for BCI

Abstract: A common problem faced by any of the devices concerning BCI applications (Brain Computer Interface) is that of the Artifacts. The undesired electrical signal interfering with the neurological activities of the brain as a consequence, misinterpreting the actual signal. The artifact may be generated internally or externally. The proposed method aims to reduce some of the predominant artifacts generated through the Electroencephalogram (EEG) such as eye blinks or artifacts generated from other devices such as Electroocculogram (EOG) or Electromyogram (EMG) . It is based on a novel combination of wavelet decomposition, Independent component analysis and thresholding. The factors for identifying the artifacts are estimated which differentiates the clean from affected EEG signal. We also intend to calculate the computational time of our proposed method for an EEG signal of 1s time frame.

Keywords: Brain Computer Interface, Artifacts, Electroencephalogram (EEG), wavelet decomposition, Independent component analysis.