Overcoming Labeled Data Barriers in Deep Ultrasound Imaging

The Medical imaging field has benefited from the application of deep networks. Early applications of deep network focused on image interpretation for the purposes of diagnosis (i.e. classification), and more recent work focuses on the use of deep networks for the image formation task (i.e. estimation). Here, we consider ultrasound image formation, and we discuss an approach to overcome the unlabeled data problem.