Abstract:SIAM Journal on Imaging Sciences, Volume 16, Issue 4, Page 2202-2246, December 2023. We introduce a method for the fast estimation of data-adapted, spatially and temporally dependent regularization parameter-maps for variational image reconstruction, focusing on total variation (TV) minimization. The proposed approach is inspired by recent developments in algorithm unrolling using deep neural networks (NNs) and relies on two distinct subnetworks. The first subnetwork estimates the regularization parameter-map from the input data. The second subnetwork unrolls [math] iterations of an iterative algorithm which approximately solves the corresponding TV-minimization problem incorporating the previously estimated regularization parameter-map. The overall network is then trained end-to-end in a supervised learning fashion using pairs of clean and corrupted data but crucially without the need for access to labels for the optimal regularization parameter-maps. We first prove consistency of the unrolled scheme by showing that the unrolled minimizing energy functional used for the supervised learning [math]-converges, as [math] tends to infinity, to the corresponding functional that incorporates the exact solution map of the TV-minimization problem. Then, we apply and evaluate the proposed method on a variety of large-scale and dynamic imaging problems with retrospectively simulated measurement data for which the automatic computation of such regularization parameters has been so far challenging using the state-of-the-art methods: a 2D dynamic cardiac magnetic resonance imaging (MRI) reconstruction problem, a quantitative brain MRI reconstruction problem, a low-dose computed tomography problem, and a dynamic image denoising problem. The proposed method consistently improves the TV reconstructions using scalar regularization parameters, and the obtained regularization parameter-maps adapt well to imaging problems and data by leading to the preservation of detailed features. Although the choice of the regularization parameter-maps is data-driven and based on NNs, the subsequent reconstruction algorithm is interpretable since it inherits the properties (e.g., convergence guarantees) of the iterative reconstruction method from which the network is implicitly defined.

Learning-Assisted Fast Determination of Regularization Parameter in Constrained Image Reconstruction.

Statistical Image Reconstruction for Low-Dose Ct Using Nonlocal Means-Based Regularization

Iteratively Refined Image Reconstruction with Learned Attentive Regularizers

Learning Weakly Convex Regularizers for Convergent Image-Reconstruction Algorithms

Deep Learning for Regularization Prediction in Diffeomorphic Image Registration

Noise-Generating and Imaging Mechanism Inspired Implicit Regularization Learning Network for Low Dose CT Reconstrution.

Conditional Deformable Image Registration with Convolutional Neural Network

Reduced iteration image reconstruction of incomplete projection CT using regularization strategy through Lp norm dictionary learning

Determining the Regularization Parameter: A Hybrid Reconstruction Technique in Fluorescence Molecular Tomography

Neural networks-based regularization for large-scale medical image reconstruction

Convergent Data-driven Regularizations for CT Reconstruction

Learning Regularization Parameter-Maps for Variational Image Reconstruction Using Deep Neural Networks and Algorithm Unrolling

Convex Latent-Optimized Adversarial Regularizers for Imaging Inverse Problems

Enhancing Low-dose CT Image Reconstruction by Integrating Supervised and Unsupervised Learning

Learned convex regularizers for inverse problems

Unsupervised Training of Convex Regularizers using Maximum Likelihood Estimation

A Neural-Network-Based Convex Regularizer for Inverse Problems

Sparse and limited-angle CT image reconstruction using dictionary learning constrained by L1 norm

Manifold constrained joint sparse learning via non-convex regularization