A Deep Cascade of Convolutional Neural Networks for Dynamic MR Image Reconstruction

By Jo Schlemper, Jose Caballero, Joseph V. Hajnal, Anthony N. Price, Daniel Rueckert. 2017

1,272 citations recorded by the source corpus.

Inspired by recent advances in deep learning, we propose a framework for reconstructing dynamic sequences of 2-D cardiac magnetic resonance (MR) images from undersampled data using a deep cascade of convolutional neural networks (CNNs) to accelerate the data acquisition process. In particular, we address the case where data are acquired using aggressive Cartesian undersampling. First, we show that when each 2-D image frame is reconstructed independently, the proposed method outperforms state-of-the-art 2-D compressed sensing approaches, such as dictionary learning-based MR image reconstruction, in terms of reconstruction error and reconstruction speed. Second, when reconstructing the frames of the sequences jointly, we demonstrate that CNNs can learn spatio-temporal correlations efficiently by combining convolution and data sharing approaches. We show that the proposed method consistently outperforms state-of-the-art methods and is capable of preserving anatomical structure more faithfully up to 11-fold undersampling. Moreover, reconstruction is very fast: each complete dynamic sequence can be reconstructed in less than 10 s and, for the 2-D case, each image frame can be reconstructed in 23 ms, enabling real-time applications.

Subfields: Biomedical Engineering, Radiology, Nuclear Medicine and Imaging

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Biomedical Engineering

In Radiology, Nuclear Medicine and Imaging

Citations in the corpus

An overview of deep learning in medical imaging focusing on MRI (2018, positive): “An open source convolutional neural networks platform for medical image analysis and image-guided therapy [320,321] http://niftynet.io DLTK. State of the art reference implementations for deep learning on medical images [131] https://github.com/DLTK/DLTK DeepMedic [322] https://github.com/Kamnitsask”
DAGAN: Deep De-Aliasing Generative Adversarial Networks for Fast Compressed Sensing MRI Reconstruction (2018, methodology): “…) = x + 0 i , and therefore the MRI forward operator can be expressed as F u combines the Fourier transform F and random undersampling U operators [51]; (2) the real and imaginary data can be learnt as separate channels in the training of the CNN, in other words, the C N is replaced as R 2 N [43].”
Deep learning in radiology: an overview of the concepts and a survey of the state of the art with focus on MRI (2018, methodology): “s were generated from 3T MR images by training a CNN with patches centered around voxels in the 3T MR images. Undersampled (in k-space) cardiac MRIs were reconstructed using a deep cascade of CNNs in [112]. A real-time method to reconstruct compressed sensed MRI using GAN was proposed by [113]. In a”
Robust Compressed Sensing MRI with Deep Generative Priors (2021, background): “[72] Jo Schlemper, Jose Caballero, Joseph V Hajnal, Anthony N Price, and Daniel Rueckert.”
Recursive Cascaded Networks for Unsupervised Medical Image Registration (2019, background): “We conduct extensive experiments based on diverse evaluation metrics (segmentations and landmarks) and multiple datasets across image types (liver CT scans and brain MRIs).”
Deep learning in medical imaging and radiation therapy. (2018, background): “Schlem-per et al. 247 evaluated the applicability of CNNs for reconstructing undersampled dynamic cardiac MR data.”
Convolutional Recurrent Neural Networks for Dynamic MR Image Reconstruction (2017, neutral): “[18] proposed a cascade network which simulated the iterative reconstruction of dictionary learning-based methods and were later extended for dynamic MR reconstructions [3].”
Deep-Learning Methods for Parallel Magnetic Resonance Imaging Reconstruction: A Survey of the Current Approaches, Trends, and Issues (2019, background): “Some architectures for image domain machine learning have been designed specifically towards a target application, for example dynamic imaging [37].”
Self‐supervised learning of physics‐guided reconstruction neural networks without fully sampled reference data (2019, methodology): “th images are obtained through SENSE-1 coil combination (2), which is the sum across the coil dimension of the product of the conjugate of the coil sensitivity maps with the corresponding coil images (31,32). Suppose that �𝑟��𝑖 is the ground truth image for subject i, and 𝑓(yΩ𝑖,𝐄Ω𝑖;𝜽) denotes”
Deep Magnetic Resonance Image Reconstruction: Inverse Problems Meet Neural Networks (2020, neutral): “Another approach of exploiting spatiotemporal correlations uses data sharing, in which the k-space data are shared among neighboring frames along the temporal direction [9].”
MoDL: Model-Based Deep Learning Architecture for Inverse Problems (2017, methodology): “An alternate approach is to unroll an iterative algorithm involving a CNN-based regularizer [16], [35], [36], which is”
NETT: solving inverse problems with deep neural networks (2018, background): “Network cascades [32, 46] and trained iterative schemes [1, 2, 12, 27] learn free parameters in iterative schemes.”
Compressed Sensing MRI Reconstruction Using a Generative Adversarial Network With a Cyclic Loss (2017, background): “[27] proposed a model that cascades a multiple deep network with data consistency layer, Lee et al.”
FastSurfer - A fast and accurate deep learning based neuroimaging pipeline (2019, neutral): “veral deep learning based approaches have been proposed for specic tasks, such as tumor segmentation [22, 23, 24, 25, 26, 27], brain lesion segmentation [28, 29, 30, 31, 32], MR image reconstruction [33, 34, 35, 36, 37, 38] or prediction of brain related diseases and their progression [39, 40, 41, 4”
Small Sample Learning in Big Data Era (2018, methodology): “Along this line of research, various unfolded technique have been successfully applied into dynamic MR image reconstruction (Schlemper et al., 2018; Qin et al., 2017), sparse view/data Computed Tomography(CT) reconstruction (Gupta et al.”