Magnetic Resonance Imaging Denoising

Magnetic resonance imaging (MRI) denoising aims to improve the quality of MRI scans by reducing noise, thereby enhancing diagnostic accuracy and potentially shortening scan times. Current research heavily utilizes deep learning, particularly employing diffusion probabilistic models, conditional neural networks, and generative adversarial networks, often in unsupervised or self-supervised frameworks to overcome the limitations of requiring high-quality ground truth data. These advancements are crucial for improving the quality of various MRI modalities, including sodium MRI and diffusion MRI, where low signal-to-noise ratios are common, ultimately leading to better clinical diagnoses and patient care. Furthermore, research is actively addressing the need for robust image quality assessment metrics, especially in the absence of reference images.