呼吸情報に着目した高精度な呼吸性洞性不整脈の抽出法

抄録

Respiratory sinus arrhythmia (RSA) is known as a selective index of cardiac vagal activity. RSA is usually extracted at a power of 0.15Hz-0.5Hz using spectral analysis. We propose a new method to extract RSA as a waveform under free breathing. First, we interpolate the RR interval (RRI) of the electrocardiogram signal by the derivative of cubic spline interpolation (DCSI), and calculate respiratory phase using Hilbert transform. Then, we extract the instantaneous RRI at every π/10 rad of the respiratory phase as the RSA waveform. Since DCSI has no delay in interpolation, instantaneous RRI keeps in phase with the variability of respiration. We extract the RSA waveform with each breath and can easily evaluate the accuracy of extraction and investigate the effect of volume or interval of respiration. Furthermore, a breath that is too short or too long in terms of the RSA frequency is removed, so it is possible to improve the accuracy of extraction. We experimented using seven subjects who conducted free breathing in resting and sitting positions for 30 minutes. Electrocardiograms and inductance plethysmographs were used at the rate of 1kHz and 100Hz, respectively, and we extracted the RSA waveform. From the analysis of mean RSA waveform, maximum point and minimum point appeared approximately at π rad and 2π rad, respectively. This result confirms that the amplitude of RSA in one breath is extracted by subtracting the RSA waveform in π rad from that in 2, π rad. From the analysis of RSA in one breath, we evaluated the accuracy of extracted RSA by S. D. in periods of 5 minutes, and obtained a value of 3.7ms. We also found that there was a positive correlation between the amplitude of RSA and the volume of respiration and interval of respiration under free breathing.