An approach for extractions of pulse and respiration information from pulsatile pressure signals

We present an approach to extract respiration signal and pulse signal that have been included in measured pulsatile signals of a pillow cuff installed on the occiput of user. In this article, we investigate a separation approach based on ICA (independent component analysis). Since there are only two observations which are two linear mixtures of three sources, pulse, respiration and noise, a usual ICA only cannot separate them perfectly. To solve this problem, we proposed a two-step algorithm; the first step includes a standard ICA and the second step includes two pieces of filter processing. In the ICA step, we can remove the noise from the mixtures of pulse and respiration signal. The filter processing can further separate the pulse and respiration. We shall present our simulation and experimental results. The obtained results are compared with reference pulse and respiration that are measured simultaneously with pillow cuff measuring, which shows effectiveness of our approach.