Artificial Intelligence in Lung Ultrasound

We summarize and review recent artificial intelligence (AI) and deep learning (DL) innovations within the field of lung point-of-care ultrasound (LUS).

Many recent DL models in LUS have been developed to aid in diagnosis of COVID-19 pneumonia, pulmonary edema, pneumothorax, and pediatric pneumonia. Publicly available data sets of patients with these pathologies have been curated, annotated, and developed to train DL models to not only detect these pathologies but also quantify severity, guide users, and act as a prognostic tool to risk stratify.

The increased use of AI in LUS has the potential to enhance medical education and to expand the field of LUS into limited resource areas such as prehospital care, disaster response, and global health. Despite the diagnostic potential of DL-enhanced LUS applications, we believe the ability of human providers to synthesize information at bedside cannot be replaced by AI.