Nonlinear Local Projection Filter for Impedance Pneumography

Abstract

The ability of impedance pneumography (IP) for recording tidal flow during long periods of free breathing make it a promising tool for assessing temporal complexity of respiration. However, techniques quantifying complexity may be sensitive to the noise in the IP signal resulting from the current processing method. A nonlinear local projection filter (NLPF) is presented as the solution to the current linear processing method, failing to reduce noise without distorting the flow signal. Current and proposed NLPF methods were applied to and existing data set of raw IP recorded in 21 infants during a methacholine challenge test. Methods’ performance was compared in a battery of test using mouth flow as a reference. NLPF achieved lower sample-by-sample error, and higher frequency attenuation, while linearity with mouth flow was maintained. Therefore, we concluded that NLPF superiorly reduces noise without distorting respiratory information.