Matthew Lungren M.D.
Assistant Professor of Radiology (Pediatric Radiology)
Darvin Yi
Graduate Student
Stanford University

ABSTRACT:

With the explosion of deep learning for computer vision, the medical imaging field is well into exploring these new tools for clinically relevant images.  However, medical image analysis poses several new problems that standard computer vision is not well equipped to solve, such as a lack of data, complicated/debatable ground truth labels, and a small inter-class variance among our different target labels.  In this talk, we present a few representative projects that explore problems in the field of classification, segmentation, and interpretation using deep learning within the medical imaging field and how these classical computer vision tasks can be optimized for medical imaging.