Adaptive spatio-temporal filtration of bioelectrical signals

In this paper we show how independent component analysis (ICA) algorithms can be used to perform spatio-temporal filtration of electromyographic (EMG) and electrocardiographic (ECG) signals. The technique was used to decompose the EMG signals into motor unit action potential (MUAP) trains. From the 88 outputs of the adaptive spatio-temporal filtration, three groups of different MUAP train patterns were found. The technique was also used to obtain a fetus' ECG and showed better result compared to using ICA.