Mathematical Foundations of AIM

With the tremendous success of deep learning in recent years, the field of medical imaging has undergone unprecedented changes. Despite the great success of deep learning in medical imaging, these recent developments are largely empirical. Our goal in this chapter is to provide an overview of some of the key mathematical foundations of deep learning to the medical imaging community. In particular, we will consider ties with traditional machine learning methods, unrolling techniques which connect deep learning to iterative algorithms, and geometric interpretations of modern deep networks.