Computational and experimental techniques towards optimising the cardiovascular risk assessment of hyperbaric decompression stress caused by circulatory bubble dynamics

Abstract

This work focuses on developing new techniques towards the quantification of hyperbaric decompression stress. Instead of just preventing decompression sickness (DCS), the aim is to go towards developing an environmental cardiovascular personalised stress index, especially as sub-clinical long terms effects of even recreational scuba diving have been demonstrated. From an engineering perspective, despite the longevity of the research field, a number of fundamental issues that remain unknown have prevented efficient modelling. The aim of this thesis is to directly tackle the research methodology by developing three tailored tools. Firstly, we develop a simulation platform in MatLab to model the diving process by optimizing the implementation of dissolved gas phase tracking decompression algorithms. This platform can be used to simulate diving scenarios, but also analyse real dive profiles. From a first analysis on real profiles provided to us by the European Divers Alert Network database, we find as expected that these existing models are poor predictors of accidents, but also demonstrate that ascent rate seems to be an important predictor of DCS for the range of profiles considered. Secondly, a fundamental issue for modelling the decompression phenomenon is that the precise formation site and growth mechanism of decompression bubbles in vivo remains unknown. We develop a novel experimental set-up and analysis code for the real-time optical study of decompression induced bubble growth dynamics. Looking at bubble growth from a gas saturated solution on ex-vivo muscle and fat tissues, we show that the role of the substrate from which bubble detach plays a significant role. Bubble density, nucleation threshold, detachment size and coalescence behaviour are shown significantly different for the two substrates, whereas growth rates after a critical size are governed by diffusion as expected, and a competition for dissolved gas between adjacent multiple bubbles is demonstrated. These findings are not accounted for in current modelling efforts so our experimental set-up could be used in the development of a more physiologically relevant decompression model. Thirdly, an important question in terms of decompression modelling optimisation is the precise definition of the evaluation endpoint. Vascular circulating bubbles are normally assessed semi-quantitatively by trained human raters who grade the severity on echocardiograms. We show statistically that this is highly rater-dependent compared to a new counting methodology which is found to perform significantly better but is more time-consuming. We then use image processing techniques to semi-automate this new counting methodology with good comparison to human raters, significantly reducing the time needed for the assessment. This new method could be added to decompression model validation protocols, as well as used in physiology experiments looking at predictive parameters for, or preventive measures against, circulating gas bubbles post-dive. The proposed experimental and computational techniques could be used towards optimising the cardiovascular risk assessment of hyperbaric decompression stress caused by circulatory bubble dynamics.