Towards Quantitative Lung Ultrasound Spectroscopy Clinical Translation

Lung Ultrasound (LUS) is a real-time and safe tool to image the lung. However, because intercostal tissues and air present large differences in acoustic impedance, LUS can only image the lung surface. This hinders the analysis to the detection of the pleural line and associated imaging artifacts, resulting in a qualitative and subjective assessment. To provide a more objective pathway, a scoring system that describes the lung conditions has been proposed, and is now the conventional LUS (C-LUS) approach clinically utilised. Although this approach is often mislabelled as quantitative LUS (Q-LUS), it remains based on subjective and qualitative image interpretation. Furthermore, C-LUS is strongly affected by both operator dependency and imaging settings, limiting its ability to provide an objective assessment of the lung and reducing both diagnostic specificity and reproducibility. Q-LUS approaches are founded on objective analysis of imaging patterns, through which biomarkers associated with physical lung characteristics can be extracted. Q-LUS might represent an important clinical improvement as it would offer objectivity and specificity to LUS analyses. Unfortunately, Q-LUS remains poorly explored in terms of reproducibility, specificity, and diagnostic potential, hindering its clinical translation. This thesis addresses these challenges by investigating the ability of Q-LUS in differentiating specific lung diseases. In addition, the diagnostic performance of Q-LUS are evaluated with two different probes, and compared with gold standard computed tomography and C-LUS. The thesis also addresses the constraint of reproducibility and subjectivity by proposing a fully automatic Q-LUS approach. Finally, a longitudinal preclinical study is conducted to investigate the relationship between Q-LUS biomarkers and different levels of lung injury in controlled experimental models, providing further validation of the potential of this technique.