Model-based filtering and compression of oscillometric blood pressure pulses

This paper presents a new approach toward outlier removal, filtering and compression of oscillometric blood pressure pulses by modeling the pulses as sum of harmonically related sinusoids. By curve fitting the proposed model to the measured oscillometric pulses using a nonlinear optimization technique, we demonstrate that an arbitrary oscillometric pulse can be modeled and consequently noise and artifacts can be reduced. As each sinusoid is precisely expressed by its amplitude, phase and frequency, the proposed method provides a compressed representation of the oscillometric pulses. We show that the proposed method achieves a compression ratio of 60 Fs/HR 2N+4, where HR is the heart rate in beats/min, Fs is the sampling frequency in Hz, and N is the number of harmonics considered in the model. New methods for detecting, replacing, and correcting the outliers based on the characteristics of the outlier neighboring pulses are also proposed in this paper.