Session S33.3

RSA Component Extraction from Heart Rate Signal by Independent Component Analysis

S Tiinanen*, M Tulppo, T Seppänen

University of Oulu
Oulu, Finland

Spectrally calculated high frequency power (HF power, 0.15-0.4Hz) of heart rate interval (RRi) series is a widely used parameter both in clinical settings and physiopathological investigations. HF power is obtained from power spectrum of RRi series by integrating the power over the HF band. The general assumption is that HF power is mainly originated from respiration, and the mechanism that affects the HF power peak is called respiratory sinus arrhythmia (RSA). Depending on the respiration rate of the subject, the RSA may overlap the low frequency (LF power, 0.04-0.15Hz) range and thus distort the frequency domain indices, e.g, the LF power or LF peak frequency. Therefore it may be useful to extract the RSA component to have “respiration-free” HRV indices. Extracted RSA component itself may also be useful index of cardiovascular system. In this paper, Independent Component Analysis (ICA) was applied to extract the RSA component from the RRi series. The ICA is a statistical method for decomposing signals into subcomponents. We used the time-synchronized respiration signal and RRi series as observation signals in ICA. The performance was evaluated with a simulation study where real 5min RRi (n=20) and respiration data (n=2) of spontaneously breathing healthy male volunteers were superimposed. According to frequency domain analysis the ICA is able to remove the RSA component without changing the power content of the RRi series (p <0.05). The residual analysis of RSA component in time domain showed that the extracted RSA component follows the shape of the original RSA component with 0.35-3.3% RMS error of total variability.

(Abstract Control Number: 149)