Study of Pulse Rate Variability Signals Using Bispectrum Analysis

In this paper, a bispectrum analysis technique is suggested for quantitative analysis of Congestive Heart Failure (CHF). The bispectrum is estimated using an autoregressive model, and the frequency coupling of the bispectrum is extracted as a quantitative measure to classify normal and CHF subjects. According to the further study of the foot to peak interval variability (FPIV), the peak to notch interval variability (PNIV) and the notch to next foot interval variability (NFIV), the experiment results showed that the PRV activity was located on a specific region of the bispectrum magnitude. Moreover, the results indicated that the frequency coupling was stable in normal subjects and messy in CHF patients. In particular, the results of the bispectrum amplitude indicated that the strength of the frequency coupling was not as strong in the normal subjects as in the CHF cases. All these changes existed in all three PRV series, especially in NFIV signal.