Breath detection algorithms effect lung function outcomes

Abstract

<b>Background:</b> Accurate breath detection is essential for the computation of outcomes in the multiple-breath washout (MBW) technique. This is particularly important in young children, where irregular breathing is common and the designation of inspirations and expirations can be challenging. <b>Aim:</b> We developed a novel algorithm for breath detection and compared it to algorithms used in commercially available MBW analysis software. <b>Methods:</b> We replicated the signal processing and algorithms used in Spiroware software (v3.2.1, Eco Medics AG). We developed custom breath detection algorithms (custom) and compared them to Spiroware using 2,424 nitrogen (N₂MBW) and 325 sulfur hexafluoride (SF₆MBW) trials collected in infants, children, and adolescents. <b>Results:</b> We found that in 2250/2424 (92.8%) N₂MBW and 311/325 (95.7%) SF₆MBW trials, the Spiroware breath detection algorithm rejected tidal breaths and did not use them for the calculation of MBW outcomes. Our custom breath detection algorithm determines inspirations and expirations based on flow reversal and corresponding CO₂ elevations, and uses all breaths for data analysis. In trials with regular tidal breathing, there were no differences in outcomes between the Spiroware and custom algorithms. However, in 14% of trials in children aged 5 to 10 years the number of breaths detected differed by ≥10% and the lung clearance index differed up to 40.5% (mean difference 3.6%). <b>Conclusion:</b> Accurate breath detection is challenging in young children. As the MBW technique relies on the cumulative analysis of all washout breaths, the rejection of breaths should be limited. Breath detection algorithms based on flow reversal and presence of CO₂ could be used as an alternative to current methods.