Introduction of a video-based heterogeneity index for quantitative lung ultrasound assessment in neonates with respiratory distress - a proof-of-concept study.

BACKGROUND: Lung ultrasound (LUS) is an established bedside tool for assessing neonatal respiratory distress. However, conventional semi-quantitative LUS scoring relies on subjective interpretation and may miss subtle but clinically relevant differences in lung aeration. We hypothesized that a novel quantitative grayscale-based Heterogeneity Index (HI) derived from LUS video loops could objectively quantify lung aeration and be associated with respiratory severity and clinical outcomes.

METHODS: In this prospective single-centre observational study, neonates with respiratory distress underwent LUS within 15-30 min after birth. LUS scoring was performed bilaterally across anterior, lateral, and posterior lung fields. Quantitative analysis used manually selected regions of interest from 2-second raw LUS video loops acquired from lateral lung fields. Grayscale- (Gray_mean, Gray_SD_time) and heterogeneity-based metrics (HI, HI_SD_time) were calculated using a custom MATLAB algorithm and correlated with oxygenation parameters, blood gas and acid-base parameters, diagnoses, and duration of respiratory support.

RESULTS: Of 47 neonates (gestational age 37.2 ± 1.8 weeks), 24 had transient tachypnoea of the newborn (TTN), 6 had respiratory distress syndrome (RDS), and 17 had delayed neonatal transition. ROC analysis demonstrated that grayscale- and HI-based metrics enabled early discrimination between respiratory diagnoses. Gray_mean and HI differentiated delayed transition from TTN (Gray_mean: AUC 0.85, 95% CI 0.72-0.96, HI: AUC 0.84, 95% CI 0.69-0.95), Gray_SD_time differentiated TTN from RDS (AUC 0.75, 95% CI 0.56-0.91), and HI_SD_time differentiated delayed transition from RDS (AUC 0.89, 95% CI 0.72-1.00). In TTN neonates, HI showed moderate discrimination for disease severity (AUC 0.72, 95% CI 0.49-0.91). Gray_SD_time correlated with FiO2, lower pH, and longer respiratory support. TTN neonates requiring prolonged respiratory support showed higher LUS scores, higher FiO2 requirements, and worse acid-base status compared to neonates with rapid resolution of symptoms. In RDS neonates, higher HI was associated with higher pCO2 and lower pH, whereas LUS scores showed no significant correlations.

CONCLUSION: Quantitative grayscale analysis of LUS video loops provides an objective measure of lung aeration. HI correlates with respiratory severity and may detect subtle changes not captured by conventional LUS scoring.