Effect of stimulating waveform and of data processing on respiratory impedance measurement

Abstract

Objective: Several commercial and custom-made forced oscillation technique (FOT) devices are used to assess respiratory system impedance. The impulse oscillometry system (IOS) is a widespread device, which yields similar but not identical results to those provided by other FOT systems. Differences may be related to the forcing waveform, the device hardware, or the data processing algorithms. We evaluated the agreement between resistance (Rrs) and reactance (Xrs) measurements while alternating between different forcing waveforms and data processing algorithms. Approach: We performed pre- and post-bronchodilator measurements in 20 patients with respiratory complaints. We generated pulse waveforms using an IOS, and sinusoidal oscillations by replacing the IOS loudspeaker with customized loudspeaker providing a 5 Hz sinusoidal pressure signal. Pressure and flow were measured using the IOS sensors and breathing circuit. We developed a data processing algorithm compatible to both forcing signals. We also applied commercial IOS software during pulse waveform and a least mean square (lms) algorithm during sinusoidal waveform. Main results: The median (5th, 95th percentile) differences between Rrs and Xrs were (1) −0.35 (−2.49, 1.23) and 0.16 (−1.63, 3.07 cmH2O*s l−1, when the same algorithm was used during pulse vs sinusoidal stimulus; (2) 0.34 (−2.33, 5.98) and 0.57 (−2.64, 6.09) cmH2O*s l−1, when our algorithm and the IOS software were used during pulse waveform; and (3) 0.33 (−1.20, 6.05) and 0.25 (−4.94, 4.28) cmH2O*s l−1 when the IOS software was used during pulse and the lms algorithm during sinusoidal waveforms. Significance: Both forcing signal and data processing contribute to differences in impedance values measured by different FOT devices.