Generalized estimation of the ventilatory distribution from the multiple-breath washout: a bench evaluation study

The multiple-breath washout (MBW) is able to provide information about the distribution of ventilation-to-volume (v/V) ratios in the lungs. However, the classical, all-parallel model may return skewed results due to the mixing effect of a common dead space. The aim of this work is to examine whether a novel mathematical model and algorithm is able to estimate v/V of a physical model, and to compare its results with those of the classical model. The novel model takes into account a dead space in series with the parallel ventilated compartments, allows for variable tidal volume (V_T) and end-expiratory lung volume (EELV), and does not require a ideal step change of the inert gas concentration.

Two physical models with preset v/V units and a common series dead space (v_d) were built and mechanically ventilated. The models underwent MBW with N₂ as inert gas, throughout which flow and N₂ concentration signals were acquired. Distribution of v/V was estimated-via nonnegative least squares, with Tikhonov regularization-with the classical, all-parallel model (with and without correction for non-ideal inspiratory N₂ step) and with the new, generalized model including breath-by-breath v_d estimates given by the Fowler method (with and without constrained V_T and EELV).

The v/V distributions estimated with constrained EELV and V_T by the generalized model were practically coincident with the actual v/V distribution for both physical models. The v/V distributions calculated with the classical model were shifted leftwards and broader as compared to the reference.

The proposed model and algorithm provided better estimates of v/V than the classical model, particularly with constrained V_T and EELV.