Keywords: Image acquisition: Reconstruction, Image acquisition: Machine learning
With their capability to push MRI towards higher resolution, higher signal-to-noise and/or shorter acquisition time, AI-based reconstruction and image enhancement techniques have quickly transitioned into clinical routine. This educational talk will illustrate how conventional MR reconstructions can be combined with trainable components and outline how these architectures can be trained for prospective use in clinical applications. Based on examples the benefits, limitations and on-going developments will be discussed.