Deep learning based dictionary learning and tomographic image reconstruction

TL;DR

This paper introduces a novel approach combining deep learning and dictionary learning for improved low-dose CT image reconstruction, leveraging sparse signal representation and modern optimization techniques.

Contribution

It presents a new interpretation of dictionary learning as a variational autoencoder and demonstrates its effectiveness in CT reconstruction.

Findings

Dictionary learning benefits from deep learning optimization methods.

Regularization by dictionaries achieves competitive CT reconstruction performance.

Sparse coding with learned dictionaries resembles a variational autoencoder.

Abstract

This work presents an approach for image reconstruction in clinical low-dose tomography that combines principles from sparse signal processing with ideas from deep learning. First, we describe sparse signal representation in terms of dictionaries from a statistical perspective and interpret dictionary learning as a process of aligning distribution that arises from a generative model with empirical distribution of true signals. As a result we can see that sparse coding with learned dictionaries resembles a specific variational autoencoder, where the decoder is a linear function and the encoder is a sparse coding algorithm. Next, we show that dictionary learning can also benefit from computational advancements introduced in the context of deep learning, such as parallelism and as stochastic optimization. Finally, we show that regularization by dictionaries achieves competitive performance in computed tomography (CT) reconstruction comparing to state-of-the-art model based and data driven approaches.