Optimal filter design to compute the mean of cardiovascular pressure signals.

Abstract

The mean pressure is a term used to describe the baseline trend of physiological pressure signals that excludes fluctuations due to the cardiac cycle and, in some cases, the respiratory cycle. In many clinical applications and bedside monitoring devices, the mean pressure is estimated with a 3-8 s moving average. We suggest that the mean pressure is best defined in terms of its frequency domain properties. This definition makes it possible to determine solutions that are both optimal and practical. We demonstrate that established methods of optimal finite impulse response (FIR) filter design produce estimates of the mean pressure that are significantly more accurate than the moving average. These filters have no more computational cost, are less sensitive to artifact, have shorter delays, and greater sensitivity to acute events.