Supervised Single Channel Source Separation Of EEG Signals
Samaneh Kouchaki, Saeid Sanei

This paper introduces a single channel source separation of electroencephalograms (EEGs) data by combining singular spectrum analysis (SSA) subspace technique and empirical mode decomposition (EMD). In the case of single channel data, many conventional techniques such as independent component analysis (ICA) cannot be directly applied. SSA is a powerful tool to analyze such data. However, the corresponding subspace of the desired signal component should be identified manually. In this work, EMD is used to supervise this procedure in places where the sources are narrowband. The results of applying the method to synthetic and real EEG data show that the supervised SSA can separate the single channel signal components automatically.