Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Deep Learning for Accelerated and Robust MRI Reconstruction: a Review

Reinhard Heckel, Mathews Jacob, Akshay Chaudhari, Or Perlman, Efrat Shimron

TL;DR

Deep learning is transforming MRI reconstruction by accelerating acquisitions and enabling robust quantitative mapping. The review surveys end-to-end networks, pretrained plug-and-play methods, generative priors, untrained networks, self-supervised learning, and modern architectures such as transformers and dual-domain networks, with a focus on robustness to distribution shifts and biases. It also covers acquisition optimization, dynamic imaging, multi-task pipelines, uncertainty estimation, and practical datasets and software, highlighting both breakthroughs and persistent challenges. The work emphasizes the need for standardized benchmarks, uncertainty quantification, and clinically informed metrics to facilitate reliable deployment in clinical practice.

Abstract

Deep learning (DL) has recently emerged as a pivotal technology for enhancing magnetic resonance imaging (MRI), a critical tool in diagnostic radiology. This review paper provides a comprehensive overview of recent advances in DL for MRI reconstruction. It focuses on DL approaches and architectures designed to improve image quality, accelerate scans, and address data-related challenges. These include end-to-end neural networks, pre-trained networks, generative models, and self-supervised methods. The paper also discusses the role of DL in optimizing acquisition protocols, enhancing robustness against distribution shifts, and tackling subtle bias. Drawing on the extensive literature and practical insights, it outlines current successes, limitations, and future directions for leveraging DL in MRI reconstruction, while emphasizing the potential of DL to significantly impact clinical imaging practices.

Deep Learning for Accelerated and Robust MRI Reconstruction: a Review

TL;DR

Deep learning is transforming MRI reconstruction by accelerating acquisitions and enabling robust quantitative mapping. The review surveys end-to-end networks, pretrained plug-and-play methods, generative priors, untrained networks, self-supervised learning, and modern architectures such as transformers and dual-domain networks, with a focus on robustness to distribution shifts and biases. It also covers acquisition optimization, dynamic imaging, multi-task pipelines, uncertainty estimation, and practical datasets and software, highlighting both breakthroughs and persistent challenges. The work emphasizes the need for standardized benchmarks, uncertainty quantification, and clinically informed metrics to facilitate reliable deployment in clinical practice.

Abstract

Deep learning (DL) has recently emerged as a pivotal technology for enhancing magnetic resonance imaging (MRI), a critical tool in diagnostic radiology. This review paper provides a comprehensive overview of recent advances in DL for MRI reconstruction. It focuses on DL approaches and architectures designed to improve image quality, accelerate scans, and address data-related challenges. These include end-to-end neural networks, pre-trained networks, generative models, and self-supervised methods. The paper also discusses the role of DL in optimizing acquisition protocols, enhancing robustness against distribution shifts, and tackling subtle bias. Drawing on the extensive literature and practical insights, it outlines current successes, limitations, and future directions for leveraging DL in MRI reconstruction, while emphasizing the potential of DL to significantly impact clinical imaging practices.
Paper Structure (46 sections, 24 equations, 4 figures)

This paper contains 46 sections, 24 equations, 4 figures.

Table of Contents

  1. Introduction
  2. Background on MRI reconstruction
  3. Image formation & forward model
  4. Conventional model-based image recovery
  5. Optimization algorithms
  6. DL reconstruction: approaches and architectures
  7. Neural networks trained end-to-end
  8. Networks mapping a noisy measurement to a clean image
  9. Unrolled network architectures
  10. Pretrained plug-and-play (PnP) methods
  11. Generative priors
  12. GANs
  13. Diffusion models
  14. Un-trained neural networks
  15. Un-trained CNNs for single image recovery
  16. ...and 31 more sections

Figures (4)

  • Figure 1: Example of the input training data for three DL reconstruction methods. The fully-supervised MoDL method aggarwal2018modl receives var-dens sampled data as input and uses the entire k-space for supervision. The self-supervised SSDU method Yaman2020 receives var-dens data as input, splits it into two subsets, and uses one set for data consistency and the other for supervision. In this example, the var-dens data were sampled from parallel-imaging (equispaced) acquired data, as in Yaman2020. The k-band method wang2023k receives var-dens sampled data from a k-space band, and uses data from the whole band for supervision, without any supervision outside the band. Different bands are acquired from different subjects, with random orientations. At inference, the input to all three methods is var-dens data from the entire k-space, similar to that shown here for MoDL.
  • Figure 2: Automated discovery of MRI acquisition protocols using supervised learning. A differentiable MR scanner utilizes the Bloch equations for in-silico signal generation and the later reconstruction of the target contrast of interest from real, acquired data. Reproduced from Loktyushin et al. loktyushin2021mrzero.
  • Figure 3: Deep learning reconstruction of quantitative magnetic resonance fingerprinting (MRF) information.(a) A fully connected neural network is trained using simulated signal trajectories. During inference, it receives a series of raw MRF images pixel-wise, as well as auxiliary maps, yielding quantitative parameter maps. (b) A further acceleration in scan time can be achieved by training a generative adversarial network (GAN) using a smaller subset of raw input data to yield the same quantitative output maps. Reproduced and modified from Weigand-Whittier et al. weigand2023accelerated.
  • Figure 4: Impact of common image perturbations on image quality metrics. A variety of image perturbations applied to a sample image from the fastMRI dataset (top row: noise addition, image blurring, pixel rolling (where an image is shifted by a number of pixels), and physics-based subject motion. The impact of these corruptions is shown for conventional image quality metrics (SSIM, PSNR) and deep feature distance metrics (LPIPS - made for natural images, SSFD - made for MR images). The deep feature metrics exhibit a larger dynamic range to the noise, blurring, and motion corruptions, but present very little change due to pixel rolling, since the image quality does not change. These qualities of deep feature metrics are ideal for assessing MRI reconstruction quality.