Unsupervised MRI Reconstruction with Generative Adversarial Networks

2020/08/29 by Cole, Elizabeth K., Pauly, John M., Vasanawala, Shreyas S. +1
#FOS: Computer and information sciences #FOS: Electrical engineering #Image and Video Processing (eess.IV) #Machine Learning (cs.LG) #Machine Learning (stat.ML) #electronic engineering #information engineering

paper · doi:10.48550/arxiv.2008.13065

Abstract

Deep learning-based image reconstruction methods have achieved promising results across multiple MRI applications. However, most approaches require large-scale fully-sampled ground truth data for supervised training. Acquiring fully-sampled data is often either difficult or impossible, particularly for dynamic contrast enhancement (DCE), 3D cardiac cine, and 4D flow. We present a deep learning framework for MRI reconstruction without any fully-sampled data using generative adversarial networks. We test the proposed method in two scenarios: retrospectively undersampled fast spin echo knee exams and prospectively undersampled abdominal DCE. The method recovers more anatomical structure compared to conventional methods.

Related

• Generalizing Deep Learning MRI Reconstruction across Different Domains
• Deep Learning for Accelerated and Robust MRI Reconstruction: a Review
• Deep MRI Reconstruction: Unrolled Optimization Algorithms Meet Neural Networks
• Deep Image Reconstruction using Unregistered Measurements without Groundtruth
• Deep Magnetic Resonance Image Reconstruction: Inverse Problems Meet Neural Networks