Radiographic and sonographic lung scores in invasively ventilated critically ill patients

Chest imaging is central to the evaluation of lung injury in invasively ventilated critically ill patients, yet the clinical utility of semiquantitative imaging scores remains uncertain. This thesis investigates the diagnostic, prognostic, and physiological relevance of the Radiographic Assessment of Lung Edema (RALE) score, the Lung Ultrasound (LUS) score, and the Chest Computed Tomography Severity Score (CTSS) in intensive care unit (ICU) patients.
Across multiple cohorts, the RALE score showed excellent diagnostic accuracy for acute respiratory distress syndrome (ARDS), with high interobserver reliability. In contrast, its prognostic performance was consistently poor, including in COVID-19 ARDS. Importantly, longitudinal changes in RALE, rather than baseline values, were independently associated with mortality, supporting a dynamic rather than static interpretation of radiographic severity.
RALE correlated weakly with CTSS, which showed a statistically significant but modest association with mortality.
Both RALE and LUS demonstrated limited correlations with gas exchange, dead space, and respiratory mechanics, indicating that imaging scores reflect only a fraction of the underlying functional impairment. In patients without ARDS, LUS failed to detect clinically meaningful changes in lung aeration in response to different levels of positive end-expiratory pressure.
Despite robust diagnostic performance, RALE has not been adopted into routine practice. Key barriers include its semiquantitative nature, need for training, and lack of evidence that its use improves clinical decision-making. Future research should move beyond diagnostic validation toward integration of RALE into multimodal, longitudinal, and potentially automated frameworks. Embedding imaging within physiological and predictive models may enable earlier risk stratification and support precision-based management of ARDS.