Unsupervised Deep Basis Pursuit: Learning inverse problems without ground-truth data

TL;DR

This paper introduces an unsupervised deep learning approach for inverse problems that learns image reconstruction without ground-truth data by jointly optimizing a neural network prior and the reconstruction process using known noise statistics.

Contribution

It proposes a novel unsupervised method that replaces traditional l1-regularization with a learned neural network prior, enabling reconstruction without ground-truth data.

Findings

Comparable performance to supervised methods

Effective in scenarios lacking ground-truth data

Potential applications in high-resolution dynamic MRI

Abstract

Basis pursuit is a compressed sensing optimization in which the l1-norm is minimized subject to model error constraints. Here we use a deep neural network prior instead of l1-regularization. Using known noise statistics, we jointly learn the prior and reconstruct images without access to ground-truth data. During training, we use alternating minimization across an unrolled iterative network and jointly solve for the neural network weights and training set image reconstructions. At inference, we fix the weights and pass the measurements through the network. We compare reconstruction performance between unsupervised and supervised (i.e. with ground-truth) methods. We hypothesize this technique could be used to learn reconstruction when ground-truth data are unavailable, such as in high-resolution dynamic MRI.