Replication Data for: Deep Learning for Detection and Localization of B-Lines in Lung Ultrasound

This dataset supports the publication: "Deep Learning for Detection and Localization of B-Lines in Lung Ultrasound." It contains 1,419 lung ultrasound videos from 113 patients admitted to the emergency department of Brigham and Women's Hospital between November 2020 and March 2021 with symptoms suggestive of a flu-like illness. All data was acquired and de-identified per institutional review board (IRB) protocols. All lung ultrasound videos were labeled for the presence of B-line artifacts. For the videos with B-lines present, annotations are provided for every fourth frame. The dataset also comprises all clip-level, frame-level, and pixel-level networks, as described in the paper.