Novel velocity model for the quantitative characterization of pleural sliding, in vivo multicenter clinical study on RF lung ultrasound data.

Quantitative lung ultrasound (LUS) has emerged as a way to improve current LUS data analysis, based on the visual interpretation of imaging patterns. However, most studies on quantitative LUS have neglected dynamic imaging patterns. Among these, pleural sliding is well-known for its clinical relevance, but few studies have focused on its quantitative characterization. Therefore, in this study, we introduce a novel methodology to estimate the velocity of pleural sliding inspired by functional ultrasound (fUS). Additionally, we assessed, through binary classifiers, the potential of this feature in distinguishing between lung pathologies. We applied the methodology to two in-vivo datasets, comprising multifrequency RF data acquired from patients affected by cardiogenic pulmonary edema (CPE), pneumonia, COPD, fibrosis, and exacerbated. Specifically, we analyzed a total of 536 multifrequency LUS videos from 79 patients. The resulting sliding velocity values were within the physiological range (≈0.49 mm/s - ≈ 4.39 mm/s). Among all pathological populations, COPD was characterized by the minimum velocity and pneumonia by the maximum velocity, on average. The statistical analysis highlighted some statistically significant differences, especially when comparing COPD with the other pathologies. These differences were confirmed also with binary classifiers that showed accuracies up to 94.06% (when differentiating between COPD and pneumonia). These results demonstrate the potential of this novel methodology to extract clinically valuable information and highlight its applicability in future diagnostic processes.