| Summary: | The ability to routinely determine Functional Residual Capacity (FRC) or the ventilation-volume homogeneity of ventilated patients has been a long held goal. Such measures have the potential to greatly improve the treatment of patients by preventing ventilator related lung injuries. Many methods have been proposed, but none have seen routine clinical use. This thesis develops methods of determining such information from tracer gas concentrations, modelling the concentration of the entire expired breath in an attempt to extract more information from the data. A method of using such models to produce a distribution of alveolar volumes, which broadens with an increase in the inhomogeneity of simulated data, was developed. When compared to the Homogeneity Index [Whiteley et al., Respir. Physiol., 124(1):65-83.], the method determines the homogeneity more robustly, particularly when noise is present in the flow signal. In real data from a water-displacement bench lung it also produced tentatively better results. However it did not perform as expected for 25% of the data sets. Investigations into determining FRC, highlighted the need to include mixing within the dead space, which is a move away from traditional dead space models. In human data a dead space with two mixing compartments provided the best FRC results, reducing the mean limits of agreement with plethysmography by 32% when compared to the other practical methods investigated. Compared to data from the literature, this method was no worse than Helium Dilution (the bias and limits of agreement were within the 95% confidence interval) and the limits of agreement were significantly better than for the LUFU device. Thus, if the results are replicated in clinical practice, the method is likely to be robust enough to positively influence patient care. Attempts to correct for time offsets between the flow and concentrations signals did not improve results for humans, despite showing some improvements for a water-displacement bench lung.
|
|---|
