Investigation on Properties of Capnogram: On Stationarity and Spectral Components of the Signal

Abstract

Capnography provides a graphical representation of the CO$_{\mathbf{2}}$ concentration in the exhaled gases. There are different methods that are used to extract time domain features of capnogram. However, those methods are manual and only suitable for normal and asthmatic capnograms. Frequency domain method is mostly used to analyze physiological signals, by assuming that those signals are stationary. Note that numerous physiological phenomena are characterized by dynamical properties. Identifying the nature of the signal is a preliminary stage which can enable to select suitable signal processing method. In this study, stationarity of capnogram signal was tested by analyzing statistical characteristics of the time series of the carbon dioxide samples recorded from normal subjects, and patients with complaint of asthma, chronic obstructive pulmonary disease, and pulmonary edema. Analysis of the spectral components of capnogram was performed using Fourier transform. The results show that there is a slight change in the statistical properties of the time series. Suggesting that capnogram can be considered as a quasi-stationary signal. Besides, analysis of capnogram based on the number of main lobes and side lobes can help to examine how changes in the spectral properties of capnogram relates to the respiratory status of the person, which can subsequentially help to discriminate normal from abnormal capnograms, and thus classify different respiratory diseases. This study may provide the key insight while identifying the proper signal processing method to analyze capnogram waveform.