AI Algorithms Can Enhance Diagnostic Performance on POCUS Studies

Background: Acute pulmonary edema is a common and serious presenting condition in the emergency department (ED). Symptoms can be variable, developing suddenly or gradually, resulting in diagnostic challenges. B-lines are discrete short path vertical reverberation artifacts (comet tail) on point-of-care ultrasound (POCUS) that manifest as echogenic lines extending from the pleural line into the lung parenchyma without diminishing in intensity and move in synchrony with lung movement. Increased lung density due to exudate, transudate, collagen, blood, etc, reduces the acoustic mismatch between the pleura and the underlying structures and results in the reflection of the ultrasound beam back to the transducer. Assessing these findings has remained a diagnostic challenge, especially for novice POCUS practitioners. Objective: To evaluate the efficacy of automatic detection of B-lines in the diagnosis of pulmonary edema using artificial intelligence (AI) (Auto B-lines). Design/Methods: Ret more...