MR image reconstruction using deep learning: evaluation of network structure and loss functions.

  title={MR image reconstruction using deep learning: evaluation of network structure and loss functions.},
  author={Vahid Ghodrati and Jiaxin Shao and Mark Bydder and Ziwu Zhou and Wotao Yin and Kim‐Lien Nguyen and Yingli Yang and Peng Hu},
  journal={Quantitative imaging in medicine and surgery},
  volume={9 9},
Background To review and evaluate approaches to convolutional neural network (CNN) reconstruction for accelerated cardiac MR imaging in the real clinical context. Methods Two CNN architectures, Unet and residual network (Resnet) were evaluated using quantitative and qualitative assessment by radiologist. Four different loss functions were also considered: pixel-wise (L1 and L2), patch-wise structural dissimilarity (Dssim) and feature-wise (perceptual loss). The networks were evaluated using… 

Figures and Tables from this paper

MR Image Reconstruction Based on Densely Connected Residual Generative Adversarial Network-DCR-GAN
A novel densely connected residual generative adversarial network (DCR-GAN) is being proposed for fast and high-quality reconstruction of MR images and it can be seen that the novel DCR-GAN leads to improved reconstruction results without a significant increase in the parameter complexity or run times.
Parallel imaging with a combination of sensitivity encoding and generative adversarial networks.
The feasibility of using GAN to improve the performance of SENSE reconstruction is demonstrated and the improvement of reconstruction is more obvious for higher under-sampling rates, which shows great potential for many clinical applications.
Prospective Deployment of Deep Learning in MRI: A Framework for Important Considerations, Challenges, and Recommendations for Best Practices
A framework for important considerations that consist of appropriate model training paradigms, evaluation of model robustness, downstream clinical utility, opportunities for future advances, as well recommendations for best current practices is provided.
DEep learning‐based rapid Spiral Image REconstruction (DESIRE) for high‐resolution spiral first‐pass myocardial perfusion imaging
It was concluded that DESIRE enabled fast and high‐quality image reconstruction for both SS and SMS MB = 2 whole‐heart high‐resolution spiral perfusion imaging.
Retrospective respiratory motion correction in cardiac cine MRI reconstruction using adversarial autoencoder and unsupervised learning
A deep neural network for respiratory motion compensation in free‐breathing cine MRI and the potential of an adversarial autoencoder network for correcting respiratory motion‐related image artifacts without requiring paired data is demonstrated.
Perfusion Maps Acquired From Dynamic Angiography MRI Using Deep Learning Approaches.
BACKGROUND A typical stroke MRI protocol includes perfusion-weighted imaging (PWI) and MR angiography (MRA), requiring a second dose of contrast agent. A deep learning method to acquire both PWI and


A Deep Cascade of Convolutional Neural Networks for Dynamic MR Image Reconstruction
A framework for reconstructing dynamic sequences of 2-D cardiac magnetic resonance images from undersampled data using a deep cascade of convolutional neural networks (CNNs) to accelerate the data acquisition process is proposed and it is demonstrated that CNNs can learn spatio-temporal correlations efficiently by combining convolution and data sharing approaches.
Deep convolutional neural networks for accelerated dynamic magnetic resonance imaging
Compared to state-of-the-art CS reconstruction techniques, this CNN achieves reconstruction speeds that are 150x faster without significant loss of image quality, and preliminary results suggest that CNNs may allow scan times that are 2x faster than those allowed by CS.
Accelerating magnetic resonance imaging via deep learning
This paper proposes a deep learning approach for accelerating magnetic resonance imaging (MRI) using a large number of existing high quality MR images as the training datasets. An off-line
Deep learning for undersampled MRI reconstruction
A deep learning method for faster magnetic resonance imaging (MRI) by reducing k-space data with sub-Nyquist sampling strategies and provides a rationale for why the proposed approach works well is presented.
Image Super-Resolution Using Deep Convolutional Networks
We propose a deep learning method for single image super-resolution (SR). Our method directly learns an end-to-end mapping between the low/high-resolution images. The mapping is represented as a deep
Convolutional Neural Networks for Inverse Problems in Imaging: A Review
Recent experimental work in convolutional neural networks to solve inverse problems in imaging, with a focus on the critical design decisions is reviewed, including sparsity-based techniques such as compressed sensing.
Deep ADMM-Net for Compressive Sensing MRI
Experiments on MRI image reconstruction under different sampling ratios in k-space demonstrate that the proposed novel ADMM-Net algorithm significantly improves the baseline ADMM algorithm and achieves high reconstruction accuracies with fast computational speed.
Natural Image Denoising with Convolutional Networks
An approach to low-level vision is presented that combines the use of convolutional networks as an image processing architecture and an unsupervised learning procedure that synthesizes training samples from specific noise models to avoid computational difficulties in MRF approaches that arise from probabilistic learning and inference.
SRPGAN: Perceptual Generative Adversarial Network for Single Image Super Resolution
A super resolution perceptual generative adversarial network (SRPGAN) framework for SISR tasks by proposing a robust perceptual loss based on the discriminator of the built SRPGAN model and combining it with the proposed perceptual loss and the adversarial loss.
Deep Convolutional Neural Network for Inverse Problems in Imaging
The proposed network outperforms total variation-regularized iterative reconstruction for the more realistic phantoms and requires less than a second to reconstruct a <inline-formula> <tex-math notation="LaTeX">$512\times 512$ </tex- math></inline- formula> image on the GPU.