Optical Coherence Tomography (OCT) is a 3D imaging technique which is widely employed by clinicians to help diagnose many eye diseases, such as macular hole and macular edema. Clinicians typically make manual measurements of macular diseases using OCT imagery for diagnosis and to guide treatment. Taking these measurements manually is time consuming and error prone. Recent advances in Deep Learning allow for detailed segmentations to be generated automatically from medical imagery, trained on images annotated by humans. In this talk, we present a form of Deep Learning known as a Convolutional Neural Network, trained to generate segmentations of macular holes and macular edema given OCT imagery. We discuss the technical and non-technical challenges and hurdles still to be overcome for Deep Learning in this field.