Current Issues in Deep Learning for Undersampled Image Reconstruction

Abstract

The subject matter is deep learning-based image reconstruction using undersampled k-space. In this literature review, I am exploring the bottlenecks that are preventing a major breakthrough in this field. I aimed to differ from the "average" literature review on this subject, which are primarily aimed at researchers. I intend for this piece to be more accessible to a broader audience. A piece of criticism I wanted to highlight in particular, are the weak quantitative metrics used in many/all papers reviewed.