Separation of artifacts from electroencephalogram signal using sequential singular spectrum analysis

Abstract

This paper presents a sequential singular spectrum analysis (SSA) also known as multistage SSA method to separate the artifacts from the single channel electroencephalogram (EEG) signal. Firstly, the (SSA) was applied on the contaminated EEG signal with window length L1□ and decomposed into three components (EOG, EEG and EMG). After observing these deco-composed components, if any artifacts are still present in the EEG components, SSA is again applied with different window length L2. Finally the artifacts such as electrooculogram (EOG) and electromyogram (EMG) are separated from the EEG signal and it is found that the seizure activity (5.45/7z)□ is preserved and all the artifact components are separated efficiently. It is also found that in terms of computational complexity the proposed sequential SSA technique is more efficient than the Local SSA.