Deep learning for fast MR imaging: A review for learning reconstruction from incomplete k-space data

Magnetic resonance imaging is a powerful imaging modality that can provide versatile information. However, it has a fundamental challenge that is time consuming to acquire images with high quality and high resolution. Reducing the scanned measurements can significantly accelerate its speed with the aid of the powerful reconstruction methods, which has evolved from linear analytic reconstructions to nonlinear iterative ones. The emerging trend in this area is replacing human-defined signal models with that learned from data. Specifically, from 2016, deep learning has been incorporated into the fast MR imaging task, which draws valuable prior knowledge from big datasets to facilitate accurate MR image reconstruction from limited measurements. Many researchers believed this started a new era of fast MR imaging techniques, namely learning reconstruction. This survey aims to review the main works in accelerating MR imaging with deep learning and will discuss merits, limitations and challenges associated with such methods. Last but not least, this paper will provide a starting point for researchers interested in contributing to this field by pointing out good tutorial resources, state-of-theart open-source codes and meaningful data sources.

Automated summary

Magnetic resonance imaging is a powerful imaging modality that can be accelerated using deep learning to provide accurate image reconstructions. By replacing human-defined signal models with models learned from data, the speed and accuracy of MR imaging techniques can be enhanced.