DAGAN: Deep De-Aliasing Generative Adversarial Networks for Fast Compressed Sensing MRI Reconstruction

Abstract

Abstract-Compressed Sensing Magnetic Resonance Imaging (CS-MRI) enables fast acquisition, which is highly desirable for numerous clinical applications. This can not only reduce the scanning cost and ease patient burden, but also potentially reduce motion artefacts and the effect of contrast washout, thus yielding better image quality. Different from parallel imaging based fast MRI, which utilises multiple coils to simultaneously receive MR signals, CS-MRI breaks the Nyquist-Shannon sampling barrier to reconstruct MRI images with much less required raw data. This paper provides a deep learning based strategy for reconstruction of CS-MRI, and bridges a substantial gap between conventional non-learning methods working only on data from a single image, and prior knowledge from large training datasets. In particular, a novel conditional Generative Adversarial Networks-based model (DAGAN) is proposed to reconstruct CS-MRI. In our DAGAN architecture, we have designed a refinement learning method to stabilise our U-Net based generator, which provides an endto-end network to reduce aliasing artefacts. To better preserve texture and edges in the reconstruction, we have coupled the adversarial loss with an innovative content loss. In addition, we incorporate frequency domain information to enforce similarity in both the image and frequency domains. We have performed comprehensive comparison studies with both conventional CS-MRI reconstruction methods and newly investigated deep learning approaches. Compared to these methods, our DAGAN method provides superior reconstruction with preserved perceptual image details. Furthermore, each image is reconstructed in about 5 ms, which is suitable for real-time processing.

Fast and accurate reconstruction of human lung gas MRI with deep learning

“…It should be noted that the AF used in the latest brain MRI reconstruction is even up to 10 . Nevertheless, the image size is 256 × 256, and the number of training data is 16,095 in that work (approximately 21 times larger than that used in this work).…”

“…However, conventional CS‐MRI had some limitations: (1) The sparse transforms (e.g., total variation or discrete wavelet transform) could be inadequate to describe complex image contents, especially for biological tissues. This could lead to artifacts and loss of detailed structures in reconstructed results . (2) CS used iterative strategies to optimize some objective functions, which was typically time‐consuming.…”

“…In particular, convolutional neural networks (CNNs) are becoming the state‐of‐the‐art strategies for image classification, super‐resolution, restoration, and segmentation . Meanwhile, CNNs based MRI reconstruction, such as variational network and deep de‐aliasing generative adversarial networks, has also been validated significant improvements over CS‐MRI regarding reconstruction quality and speed . Such methods learn the nonlinear mapping between the undersampled images or k‐space and reference images by training CNNs .…”

Reconstruction of multicontrast MR images through deep learning

“…When the acceleration factor was 8, both outputs from X‐net and Y‐net were slightly blurred. It has been demonstrated that this type of blurring effect can be offset by utilizing an adversarial network …”

Deep residual network for off‐resonance artifact correction with application to pediatric body MRA with 3D cones

“…There have been many published techniques on undersampled image reconstruction, and image reconstruction can also be interpreted as an image artifact correction problem, suggesting feasibility for deblurring off‐resonance. Some convolutional neural network techniques operate entirely in the image domain and enhance image quality with supervised, perceptual, and adversarial losses . Still operating primarily in the image domain but drawing from a SENSE‐based reconstruction, there are techniques that use deep variational networks to learn a deep model prior to replacing the sparsity regularizers in the compressed‐sensing reconstruction equation .…”