De-Aliasing and Accelerated Sparse Magnetic Resonance Image Reconstruction Using Fully Dense CNN with Attention Gates

Abstract

When sparsely sampled data are used to accelerate magnetic resonance imaging (MRI), conventional reconstruction approaches produce significant artifacts that obscure the content of the image. To remove aliasing artifacts, we propose an advanced convolutional neural network (CNN) called fully dense attention CNN (FDA-CNN). We updated the Unet model with the fully dense connectivity and attention mechanism for MRI reconstruction. The main benefit of FDA-CNN is that an attention gate in each decoder layer increases the learning process by focusing on the relevant image features and provides a better generalization of the network by reducing irrelevant activations. Moreover, densely interconnected convolutional layers reuse the feature maps and prevent the vanishing gradient problem. Additionally, we also implement a new, proficient under-sampling pattern in the phase direction that takes low and high frequencies from the k-space both randomly and non-randomly. The performance of FDA-CNN was evaluated quantitatively and qualitatively with three different sub-sampling masks and datasets. Compared with five current deep learning-based and two compressed sensing MRI reconstruction techniques, the proposed method performed better as it reconstructed smoother and brighter images. Furthermore, FDA-CNN improved the mean PSNR by 2 dB, SSIM by 0.35, and VIFP by 0.37 compared with Unet for the acceleration factor of 5.

A Hybrid Residual Attention Convolutional Neural Network for Compressed Sensing Magnetic Resonance Image Reconstruction

“…The effectiveness of our proposed RA-CNN was compared with direct mapping and single- and multi-domain networks. Unet [ 36 ], the de-aliasing generative adversarial network (DAGAN) [ 13 ], RefineGAN [ 14 ], the projection-based cascade Unet (PBCU) [ 51 ], and the fully dense attention (FDA)-CNN [ 52 ] are single-domain techniques, while the DC-CNN [ 34 ], KIKI-net [ 29 ], W-net [ 35 ], the hybrid cascade [ 30 ], and the dual-encoder Unet [ 53 ] are multi-domain approaches. The GAN-based DAGAN applies a residual Unet architecture for the generator and combined adversarial and innovative content losses.…”

Emerging Trends in Fast MRI Using Deep-Learning Reconstruction on Undersampled k-Space Data: A Systematic Review

“…Although attention modules can introduce computational demands, memory-efficient self-attention modules have been developed to address this limitation, making the integration of attention mechanisms more efficient in MRI reconstruction [ 88 , 89 ]. Overall, the attention module efficiently guides the network’s attention to relevant image content, contributing to MRI reconstruction [ 90 , 91 , 92 , 93 , 94 , 95 , 96 ].…”

AI in MRI: Computational Frameworks for a Faster, Optimized, and Automated Imaging Workflow

“…Retrospective experiments with 2D and 3D MRI data indicate that the PROJECTOR-generated trajectories can exploit the full possible range of gradients and slew rates well and produce sharp images. In another work, Hossain et al [ 46 ] propose a new sampling pattern for 2D MRI, which combines the random and non-random sampling of the phase-encoding direction. The authors also introduce an advanced fully dense attention convolutional neural network (FDA-CNN), which reduces the number of redundant features using attention gates.…”