MRI Banding Removal via Adversarial Training

MR images reconstructed from sub-sampled Cartesian data using deep learning techniques show a characteristic banding (sometimes described as streaking), which is particularly strong in low signal-to-noise regions of the reconstructed image. These unnatural artifacts have been identified as one of the largest obstacles preventing the clinical use of machine-learning based MRI reconstructions. In this work, we propose the use of an adversarial loss that penalizes banding structures without requiring any human annotation. Our technique greatly reduces the appearance of banding, without requiring any additional computation or post-processing at reconstruction time. Our approach is compatible with any existing reconstruction approach that uses supervised machine learning, including the current state-ofthe-art. We report the results of a blind comparison against a strong baseline by a group of expert evaluators (board-certified radiologists), where our approach is ranked superior at banding removal with no statistically significant loss of detail. A reference implementation of our method is available in the supplementary material.