ICA-Derived Respiration Using an Adaptive R-peak Detector

Abstract

Breathing Rate (BR) plays a key role in health deterioration monitoring. Despite that, it has been neglected due to inadequate nursing skills and insufficient equipment. ECG signal, which is always monitored in a hospital ward, is affected by respiration which makes it a highly appealing way for the BR estimation. In addition, the latter requires accurate R-peak detection, which is a continuing concern because current methods are still inaccurate and miss heart beats. This study proposes a frequency domain BR estimation method which uses a novel real-time R-peak detector based on Empirical Mode Decomposition (EMD) and a blind source ICA for separating the respiratory signal. The originality of the BR estimation method is that it takes place in the frequency domain as opposed to some of the current methods which rely on a time domain analysis, making the estimation more accurate. Moreover, our novel QRS detector uses an adaptive threshold over a sliding window and differentiates large Q-peaks from R-peaks, facilitating a more accurate BR estimation. The performance of our methods was tested on real data from Capnobase dataset. An average mean absolute error of less than 0.7 breath per minute was achieved using our frequency domain method, compared to 15 breaths per minute of the time domain analysis. Moreover, our modified QRS detector shows comparable results to other published methods, achieving a detection rate over 99.80%.