Principal component analysis tensor decomposition method to remove ocular artifact

Electroencephalogram (EEG) is easily polluted by other biomedical signals that influence the disease diagnosis. The waveform of ocular artifacts is similar with epilepsy. It is a significant problem to remove ocular artifacts. At present, the independent component analysis (ICA) is used widely to remove ocular artifacts. However, the ICA is usually used to resolve the problem when the number of source equals the number of observed signals. So we proposed a principal component analysis tensor decomposition method to solve the problem of underdetermined blind source separation. The simulations show that this method is better than the ICA.