Automatic Removal of Eye Blink Artefacts from EEG Data Using Spatio-Temporal Features

Abstract

Electroencephalography (EEG) is a standard method in which electrical signals of cerebral activities are collected using electrodes set along the scalp. It is a non-invasive technique used for clinical applications. The common problem with EEG data is that it is susceptible to the imitation of the distinct biological or environmental noise interferences well known as artefacts. Researchers have proposed various methods to eliminate different types of artefacts from the contaminated EEG signal, yet nothing stands standard for endorsement of recorded EEG signals in clinical usage. Consequently, exploration of artefacts removal research remains engaging and challenging even today. In this paper, simulations are performed using a proficient fast independent component analysis (ICA)-based automatic EEG artifact detection based on joint use of spatial and temporal features (ADJUST) method to remove ocular artefacts from EEG signal using the spatio-temporal features. The proposed algorithm has the major function of signal disintegration, temporal and spatial features computation, and classification. This algorithm differentiates artefacts from the independent components segment. The proposed method follows the binary classification of the artefact or non-artefact present in the EEG signal in the pre-final stage while in the subsequent stage, the artefact correction is done. Performance of this algorithm is compared with another well-known ICA algorithm (ARA, i.e. artefact removing algorithm) and the proposed algorithm is producing 18% better outcomes than ARA. Hence, this proposed idea removes the artefacts from the contaminated EEG signal effectively, which might help the neurologists acquire the right information from EEG and enable proper clinical diagnosis.