Recurrent Generative Adversarial Networks for Proximal Learning and Automated Compressive Image Recovery

Abstract

Recovering images from undersampled linear measurements typically leads to an ill-posed linear inverse problem, that asks for proper statistical priors. Building effective priors is however challenged by the low train and test overhead dictated by real-time tasks; and the need for retrieving visually "plausible" and physically "feasible" images with minimal hallucination. To cope with these challenges, we design a cascaded network architecture that unrolls the proximal gradient iterations by permeating benefits from generative residual networks (ResNet) to modeling the proximal operator. A mixture of pixel-wise and perceptual costs is then deployed to train proximals. The overall architecture resembles back-and-forth projection onto the intersection of feasible and plausible images. Extensive computational experiments are examined for a global task of reconstructing MR images of pediatric patients, and a more local task of superresolving CelebA faces, that are insightful to design efficient architectures. Our observations indicate that for MRI reconstruction, a recurrent ResNet with a single residual block effectively learns the proximal. This simple architecture appears to significantly outperform the alternative deep ResNet architecture by 2dB SNR, and the conventional compressed-sensing MRI by 4dB SNR with 100x faster inference. For image superresolution, our preliminary results indicate that modeling the denoising proximal demands deep ResNets.

11 pages, 11 figures

References in corpus (8)

• Low-Dose CT with a Residual Encoder-Decoder Convolutional Neural Network (RED-CNN)
• Perceptual Losses for Real-Time Style Transfer and Super-Resolution
• Least Squares Generative Adversarial Networks
• Deep Generative Adversarial Networks for Compressed Sensing Automates MRI
• Deep Learning Face Attributes in the Wild
• Super-Resolution with Deep Convolutional Sufficient Statistics
• Learning to solve inverse problems using Wasserstein loss
• Real-time Dynamic MRI Reconstruction using Stacked Denoising Autoencoder

Cited by in corpus (18)

• Recent Advances in Recurrent Neural Networks
• MR image reconstruction using deep density priors
• Compressed Sensing with Deep Image Prior and Learned Regularization
• Dense Recurrent Neural Networks for Accelerated MRI: History-Cognizant Unrolling of Optimization Algorithms
• High-Resolution Limited-Angle Phase Tomography of Dense Layered Objects Using Deep Neural Networks
• Low-Dose CT with Deep Learning Regularization via Proximal Forward Backward Splitting
• Task adapted reconstruction for inverse problems
• Convolutional Recurrent Neural Networks for Dynamic MR Image Reconstruction
• Spectral pre-modulation of training examples enhances the spatial resolution of the Phase Extraction Neural Network (PhENN)
• Limited-angle tomographic reconstruction of dense layered objects by dynamical machine learning
• Unified Supervised-Unsupervised (SUPER) Learning for X-ray CT Image Reconstruction
• A Provably Convergent Scheme for Compressive Sensing under Random Generative Priors
• DeepcomplexMRI: Exploiting deep residual network for fast parallel MR imaging with complex convolution
• Big in Japan: Regularizing networks for solving inverse problems
• Learning to synthesize: splitting and recombining low and high spatial frequencies for image recovery
• Synergistic Reconstruction and Synthesis via Generative Adversarial Networks for Accelerated Multi-Contrast MRI
• Deep Unfolding of a Proximal Interior Point Method for Image Restoration
• AHP-Net: adaptive-hyper-parameter deep learning based image reconstruction method for multilevel low-dose CT