Improving Limited Angle CT Reconstruction with a Robust GAN Prior

TL;DR

This paper introduces a robust GAN-based prior for limited angle CT reconstruction, effectively reducing artifacts and noise without needing training or measurement model access, thus enhancing image quality in various scenarios.

Contribution

The work presents a novel robust GAN prior based on corruption mimicking for improved limited angle CT reconstruction, operating directly in image space without additional training.

Findings

Significantly improves reconstruction quality with fewer artifacts.

Operates scanner-agnostic and does not require measurement model access.

Effective across diverse sensing scenarios.

Abstract

Limited angle CT reconstruction is an under-determined linear inverse problem that requires appropriate regularization techniques to be solved. In this work we study how pre-trained generative adversarial networks (GANs) can be used to clean noisy, highly artifact laden reconstructions from conventional techniques, by effectively projecting onto the inferred image manifold. In particular, we use a robust version of the popularly used GAN prior for inverse problems, based on a recent technique called corruption mimicking, that significantly improves the reconstruction quality. The proposed approach operates in the image space directly, as a result of which it does not need to be trained or require access to the measurement model, is scanner agnostic, and can work over a wide range of sensing scenarios.