Background: Pendelluft results from differences in time constants, airway opening pressures and heterogenous pressure distributions. It has been associated with increased lung strain, and worse clinical outcomes (4). Although scientific interest in pendelluft has grown, no consensus exists on the optimal method for its detection and quantification. This thesis aimed to compare five EIT-based pendelluft quantification methods regarding pendelluft prevalence across different datasets of mechanically ventilated ICU patients.

Methods: The five methods (5-9) were implemented as described in their original publications and applied to three patient groups: those with healthy lungs, those with hypoxemic failure, and those on non-invasive ventilation. To compare prevalence results, thresholds were derived from measurements in patients on pressure-controlled ventilation. Each method was evaluated using both its original preprocessing, and a standardized preprocessing approach to assess the impact of preprocessing choices.

Results: Pendelluft prevalence was highest among hypoxemic failure patients (up to 100%), and lowest in patients with healthy lungs (as low as 20%), with agreement across methods ranging from 62-83%. Several methods showed significant positive correlations. Pendelluft intensity was significantly higher during pressure-support ventilation compared to pressure-controlled ventilation (n=15, p = 0.03).

Discussion: While agreement on pendelluft prevalence was relatively high, differences in intensity reflected underlying methodological variations in signal interpretation. Standardizing preprocessing improved comparability and reduced variability. A significant increase in pendelluft during pressure-support ventilation supports its link to patient effort. These findings highlight the need for standardized quantification protocols and emphasize the impact of preprocessing choices in advancing pendelluft research and clinical applications.