This event is for the Stanford community.
Magnetic resonance imaging is a powerful diagnostic imaging modality, however, generating high-resolution MR images is challenging due to constraints of long scan times and low signal to noise ratios. In this presentation, I will describe the use of deep-learning based super-resolution techniques to enhance the resolution of low-resolution MRI scans. I will describe the algorithms used for super-resolution and present frameworks that could be used to evaluate the true efficacy of the technique, beyond simple metrics of image quality enhancement used in traditional computer vision. I will end with a brief description on how these techniques could be used for simultaneous and end-to-end optimization of upstream image acquisition and downstream image analysis.