Abstract:Recently, fast Magnetic Resonance Imaging reconstruction technology has emerged as a promising way to improve the clinical diagnostic experience by significantly reducing scan times. While existing studies have used Generative Adversarial Networks to achieve impressive results in reconstructing MR images, they still suffer from challenges such as blurred zones/boundaries and abnormal spots caused by inevitable noise in the reconstruction process. To this end, we propose a novel deep framework termed Anisotropic Diffusion-Assisted Generative Adversarial Networks, which aims to maximally preserve valid high-frequency information and structural details while minimizing noises in reconstructed images by optimizing a joint loss function in a unified framework. In doing so, it enables more authentic and accurate MR image generation. To specifically handle unforeseeable noises, an Anisotropic Diffused Reconstruction Module is developed and added aside the backbone network as a denoise assistant, which improves the final image quality by minimizing reconstruction losses between targets and iteratively denoised generative outputs with no extra computational complexity during the testing phase. To make the most of valuable MRI data, we extend its application to support multi-modal learning to boost reconstructed image quality by aggregating more valid information from images of diverse modalities. Extensive experiments on public datasets show that the proposed framework can achieve superior performance in polishing up the quality of reconstructed MR images. For example, the proposed method obtains average PSNR and mSSIM values of 35.785dB and 0.9765 on the MRNet dataset, which are at least about 2.9dB and 0.07 higher than those from the baselines.

Analysis and Research on the Generating Mechanism and Solutions of Gradient Field Artifacts in MR Images

Generating Mechanism and Solutions of MRI Radio Frequency Field Artifacts

Characterization and Reduction of X-Gradient Induced Eddy Currents in A Ndfeb Magnetic Resonance Imaging Magnet-3d Finite Element Method-Based Numerical Studies

Reduction of artifacts in susceptibility-weighted MR venography of the brain.

Research of Generation Mechanism and Compensation Techniques of MRI Artifacts

Unconventional Gradient Coil Designs in Magnetic Resonance Imaging.

Finite Element Analysis of Gradient Z-Coil Induced Eddy Currents in a Permanent MRI Magnet

Analysis of Influencing Factors and Preventive Measures for RF Field Artifacts in MR Images

Influence of Dental Metallic Materials on MR Imaging

A Structure Design Method for Reduction of MRI Acoustic Noise.

Gradient Coil Design Method Specifically for Permanent-Magnet-Type Low Field Portable MRI Brain Scanner

Towards High-Quality MRI Reconstruction With Anisotropic Diffusion-Assisted Generative Adversarial Networks And Its Multi-Modal Images Extension

Z-Gradient Coil Design with Improved Anti-eddy Performance for MRI System with Opposed Permanent Magnets

Multi-Contrast Brain MRI Image Super-Resolution with Gradient-Guided Edge Enhancement.

An Adaptive Anisotropic Diffusion Filter for Noise Reduction in MR Images

Analysis and Research of Factors Influencing the Quality of EBCT Images

A Simple Enhancement Algorithm for MR Head Images

EG<SUP>2</SUP>N: Enhanced Gradient Guiding Network for Single MR Image Super-Resolution

Design of High Performance DSP-based Gradient Calculation Module for MRI

Designing Gradient Coils Considering the Ferromagnetic Effect for a Movable Permanent MRI System

Deep Learning for Image Enhancement and Correction in Magnetic Resonance Imaging—State-of-the-Art and Challenges