Performance of the EDR Methods: Evaluations Using the Mean and Instantaneous Respiration Rates

Abstract

Performance of the 6 different ECG-derived respiration (EDR) methods is evaluated in this study. Data obtained from the nasal, abdominal and chest based respiratory signals and derived from simultaneously recorded ECG signals of the polysomnography recordings related to the 5 different patients are used for evaluations. To determine the estimated instantaneous respiratory rates (IRRs) and corresponding measuring times along the 10 minutes long respiratory signals, a peak detection algorithm measuring the time interval between the sequential peaks are used. Mean respiratory rates (MRRs) are determined by using the time series of estimated IRRs related to the observed and derived respiratory signals. However, the homogeneous time series of IRRs produced by a process of 1 Hz re-sampling following the cubic interpolation are used for calculation of the correlations. Evaluations based on the assessment of MRR and correlation of the homogenous IRR signals showed that the EDR methods using the tools related to the peak-to-peak interval measurements and band-pass filtering of the ECG signal has the best performance referring to the nasal based respiratory signal. It is also observed that abdominal and chest based estimation of the IRRs can be given as a reference respiratory signal, with 70% approximate accuracy, for evaluation of EDR methods.