RSA component extraction from cardiovascular signals by combining adaptive filtering and PCA derived respiration

Respiratory sinus arrhythmia (RSA) means heart rate changing synchronously with respiration and is usually in high frequency range (HF, 0.15-0.4Hz). Depending on measurement protocol, respiration rates may alter in both low frequency (LF, 0.04-0.15Hz) and HF range distorting frequency domain indices of heart rate interval (RRi) series and systolic blood pressures (SBP) series. Adaptive filtering can be used to extract the RSA component from cardiovascular signals. However, this method requires a reference respiration signal. We demonstrate how ECG derived surrogate respiration by principal component analysis (PCA) can be used as a reference signal in Least Mean Square (LMS) adaptive filter. Data set consist of 23 healthy males performing spontaneous breathing at rest. RRi and SBP series were adaptively filtered using measured respiration and ECG derived respiration. We conclude that the ECG-based respiration surrogate is adequate to extract the RSA component.