MRI at 0.55T: Opportunities, Challenges, and Recent Advances

Low-field MRI at 0.55T has emerged as a promising platform for both research and clinical use in recent years, offering unique advantages that include lower system cost, reduced susceptibility-related artifacts, and greater flexibility in system and acquisition design. At the same time, a fundamental limitation of MRI at 0.55T is the inherently lower signal-to-noise ratio (SNR), which limits achievable spatial and temporal resolution and often requires prolonged scan times for many applications. This talk will describe recent advances at NYU that aim to address these challenges through a combination of acquisition strategies and image reconstruction methods, with a particular emphasis on self-supervised learning-based image denoising and reconstruction. Unlike conventional supervised approaches, which rely on high-SNR reference images that are often unavailable at low field, self-supervised strategies exploit statistical properties of noise and data redundancy to generate clean images directly from routinely acquired noisy data. Following this, the talk will also discuss future directions for integrating deep learning methods into 0.55T MRI workflows, and how these advances may help unlock the full potential of low-field MRI for clinical and translational imaging.

Dr. Li Feng is an Associate Professor of Radiology at the New York University Grossman School of Medicine, where he serves as Director of Rapid Imaging and oversees research efforts in rapid MRI within the Department of Radiology. He is an internationally recognized expert in rapid MRI, non-Cartesian MRI, and free-breathing MRI, and has pioneered several fast-imaging methods that have been successfully translated into clinical practice. In recent years, Dr. Feng’s research has also focused on MRI at 0.55T, with particular emphasis on developing self-supervised-based approaches to improve image quality at low field.